AI News Pulse

Subscriptions

Every way to buy AI

Subscriptions, coding plans, bundles, gateways and APIs, with how each plan's limits really work and the cheapest way to run a model. Every entry is checked against the provider's own page and dated.

62 plans from 17 providers, checked 2026-10-21

Coding plan

36 entries, cheapest first

  • Cursor

    Free

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $0
    per month
    Limits
    The Free plan's usage is labelled 'Limited' with no dollar or request amount. It includes Agent, Chat and Tab completions with Cursor models (Grok and Composer), per the Cursor models and pricing docs. The page does not list the exact models. usage is described only as 'Limited'; no dollars, requests or tokens are published for it.
    Rules
    • Plus and Free do not include third-party models
    Works with
    • Cursor
    More details
    • Part of The X Subscription lineup on Cursor's current pages.
    Verified 21 Oct 2026Source ↗Pricing ↗Buy on the official site
  • Meta

    Everyday Usage

    Coding planMessagesOpaque: no published token or dollar conversion
    $5
    per month
    Limits
    Everyday Usage allows 10-50 prompts every 5 hours, including image and video uploads. At the limit you can upgrade to the next plan or wait for the limit to refresh. Subscriptions work only in the Muse Code CLI. the allowance is a prompt range (10-50 per 5 hours) with no published token or dollar conversion for subscriptions; API rates are published separately.
    Rules
    • Works only through the Muse Code CLI while signed in with your Meta Model API account; other API keys you create are billed pay-as-you-go
    • Must be 18 or the age of majority in your country
    • Benefits and availability vary by region; Muse Wonder and Muse Wonder Full are not available in South Korea and Muse Image is not available in India
    Models
    Works with
    • Muse Code CLI
    More details
    • Voice mode and web search are included
    • Subscriptions default to Contributor mode: more usage in exchange for Meta using your content to train future models; you can turn it off at any time
    • Upgrades take effect immediately with a prorated charge; downgrades start in the next billing cycle; cancellations stay active until the cycle ends
    • Tier names and prices are from the Muse Code product page. The subscriptions doc does not print prices and says benefits are shown during onboarding.
    Verified 21 Oct 2026Source ↗Muse Code subscriptions ↗What is a Muse Code subscription ↗Buy on the official site
  • Xiaomi (MiMo)

    MiMo Token Plan Lite

    Coding planCreditsClear quotaOff-peak 00:00-08:00 UTC+8 (Beijing time; 16:00-24:00 UTC): Consumption coefficient 0.8x (20% fewer credits)
    ¥39
    per month
    or $6 per month
    or ¥408 per year (¥34/mo)
    or $63.36 per year ($5.28/mo)
    Limits
    4,100,000,000 credits per month
    A fixed monthly quota of 4.1 billion credits, converted to tokens at the published per-model rates. When the quota runs out the service pauses; it does not draw on bonus or account balance. Unused quota does not carry over, and no 5-hour or weekly cap is listed.
    Rules
    • Quota may only be used in programming tools such as OpenClaw and OpenCode (Token Plan subscription page)
    • Prohibited: API calls for clearly non-coding scenarios, such as automated scripts and custom application backends; violations may lead to suspension of the subscription and blocking of the API key (Token Plan subscription page)
    Models
    Works with
    • Claude Code
    • OpenCode
    • OpenClaw
    Worth knowing
    Off-peak: 00:00-08:00 (UTC+8 (Beijing time; 16:00-24:00 UTC)), Consumption coefficient 0.8x (20% fewer credits).
    More details
    • First purchase: 12% off, once per account (Individual Edition); not applicable to annual auto-renewal
    • TTS models are free for a limited time and do not consume plan credits
    • Mimo-v2.5-pro and mimo-v2.5 retire at 10:00 Beijing time on October 21, 2026, per the price page; only mimo-v2.6 models are listed. A Team Edition sells Standard, Pro and Max seats at the same prices per seat; it is not recorded separately. The plan page also lists Xiaomi's speech recognition (ASR) and text to speech (TTS) models.
    Verified 21 Oct 2026Source ↗Token Plan subscription terms ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Lite

    Coding planInternational (Singapore region, USD)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash (page: 40% of normal cost); 50% off for deepseek-v4-pro-0813, deepseek-v4-flash-0731, deepseek-v4.1-flash and deepseek-v4.1-flash-1017
    $6
    per month
    Limited-time price, regular $8
    or $65 per year ($5.42/mo)
    Limits
    11,500 credits per month
    A monthly quota of 11,500 credits; the weekly cap was removed on September 22, 2026. Credits per request depend on the model, token usage, thinking mode and tool calls, and the console shows per-call usage. 1-2 concurrent agents. Credits; Alibaba publishes no credit-to-token rate (the page says credits are dynamically determined by model type, token usage, thinking mode and tool calls).
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan overview)
    • Available only in the Singapore region with Global deployment: inference runs globally and prompts and outputs cross borders (Token Plan overview)
    • For the subscriber's exclusive use; sharing the key with others may lead to suspension (Token Plan overview)
    • The Personal Edition does not support unsubscription; an expired subscription disables the API key (Token Plan overview)
    Models
    and 12 more
    • wan2.7-image-pro
    • wan3.0-video
    • deepseek-v4.1-flash
    • deepseek-v4.1-flash-1017
    • deepseek-v4-pro
    • deepseek-v4-pro-0813
    • deepseek-v4-flash-0731
    • glm-5.3
    • glm-5.2
    • happyhorse-1.1-i2v
    • happyhorse-1.1-t2v
    • happyhorse-1.1-r2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash (page: 40% of normal cost); 50% off for deepseek-v4-pro-0813, deepseek-v4-flash-0731, deepseek-v4.1-flash and deepseek-v4.1-flash-1017.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Concurrent agents: 1-2 (Lite), 2-3 (Essential), 3-4 (Standard), 6-8 (Pro)
    • Extra Bundle: $15 per bundle per month for 20,000 Credits; up to 5 held at once; needs an active subscription
    • Upgrades are prorated: you pay the price difference for the remaining days and get prorated extra credits
    • One subscription per verified identity per edition
    • The international edition of the Token Plan; the China edition (Beijing region, CNY) is a separate entry with the same credit quotas.
    Note
    Verified 21 Oct 2026Source ↗Token Plan landing page (prices) ↗Buy on the official site
  • Cursor

    Plus

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $8
    per month
    or $80 per year ($6.67/mo)
    Limits
    Paid plans include a weekly pool of usage for Cursor models, shared across Grok and Cursor. Different models draw from the pool at different rates. The pool is described only as a multiple of Plus ('Base'); no dollar or request amount is published. Cursor publishes per-token rates for usage beyond the pool. pool sizes are published only as multiples of Plus ('Base', '4x Plus', '15x Plus', '50x Plus'), with no dollar or request conversion.
    Rules
    • Auto, Cloud Agents, Automations, Bugbot, Origin and third-party models are not included
    • Prices exclude tax. Cancel at least 24 hours before renewal to avoid additional charges
    Models
    Works with
    • Cursor
    More details
    • When the weekly pool runs out you can wait for the reset, use prepaid credit, or upgrade; requests are not downgraded in quality or speed
    • Top-ups add prepaid credit; Auto Top-up refills the balance below a threshold you set; a monthly additional usage limit caps paid additional usage
    • The pricing page footnote says only Grok models are on this plan; the models docs list Composer 2.5 alongside the Grok models for Plus, at fixed settings (Fast off, medium effort). Legacy plans Pro ($20/month), Pro+ ($60/month) and the legacy Ultra ($200/month) are not recorded here: the legacy page says new customers start on The X Subscription.
    Verified 21 Oct 2026Source ↗Pricing ↗Pricing and plans help ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Essential

    Coding planInternational (Singapore region, USD)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models
    $10
    per month
    Limited-time price, regular $16
    or $100 per year ($8.33/mo)
    Limits
    25,500 credits per month
    A monthly quota of 25,500 credits; the weekly cap was removed on September 22, 2026. Credits per request depend on the model, token usage, thinking mode and tool calls, and the console shows per-call usage. 2-3 concurrent agents. Credits; Alibaba publishes no credit-to-token rate.
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan overview)
    • Available only in the Singapore region with Global deployment: inference runs globally and prompts and outputs cross borders (Token Plan overview)
    • For the subscriber's exclusive use; sharing the key with others may lead to suspension (Token Plan overview)
    • The Personal Edition does not support unsubscription; an expired subscription disables the API key (Token Plan overview)
    Models
    and 12 more
    • wan2.7-image-pro
    • wan3.0-video
    • deepseek-v4.1-flash
    • deepseek-v4.1-flash-1017
    • deepseek-v4-pro
    • deepseek-v4-pro-0813
    • deepseek-v4-flash-0731
    • glm-5.3
    • glm-5.2
    • happyhorse-1.1-i2v
    • happyhorse-1.1-t2v
    • happyhorse-1.1-r2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Concurrent agents: 1-2 (Lite), 2-3 (Essential), 3-4 (Standard), 6-8 (Pro)
    • Extra Bundle: $15 per bundle per month for 20,000 Credits; up to 5 held at once; needs an active subscription
    • Upgrades are prorated: you pay the price difference for the remaining days and get prorated extra credits
    • One subscription per verified identity per edition
    • The international edition of the Token Plan; the China edition (Beijing region, CNY) is a separate entry with the same credit quotas.
    Note
    Verified 21 Oct 2026Source ↗Token Plan landing page (prices) ↗Buy on the official site
  • OpenCode

    Go

    Coding planDollar allowanceClear quota
    $10
    per month
    Limits
    20% of monthly limit per 5 hours50% of monthly limit per week100% of monthly limit per month
    Usage limits are defined as monthly dollar amounts, set per model. A model's 5-hour limit is 20% of its monthly limit, its weekly limit is 50% and its monthly limit is 100%. The FAQ says these quotas are enforced across rolling windows. Request counts shown per model are estimates.
    Rules
    • Usage limits may change as OpenCode learns from early usage and feedback
    • Cancel any time
    Models
    and 19 more
    • GLM-5.3
    • GLM-5.2
    • Kimi K3
    • Kimi K2.7 Code
    • LongCat-2.0
    • MiMo-V2.6-Pro
    • MiMo-V2.6-Flash
    • MiMo-V2.5-Pro
    • MiMo-V2.5
    • Qwen3.8 Max
    • Qwen3.8 Flash
    • Qwen3.7 Plus
    • MiniMax M3
    • Muse Spark 1.3 Contributor
    • Muse Spark 1.2 Contributor
    • DeepSeek V4.1 Flash
    • DeepSeek V4 Pro
    • Hy4 preview
    • Hy3
    Works with
    • OpenCode
    More details
    • Extra usage: turn it on in the Console Overview credits card; after the limits are reached, Go falls back to the pay-as-you-go balance instead of blocking requests. Without Extra usage, the free models remain available
    • Setup guide names Claude Code, Codex, Hermes, Pi and Raycast as agents to set up with the Go key; the page says 'Use with any agent'
    • Refund possible if the charge was made within the last 14 days and the Go allowance is unused in that billing period
    • Limited-time multipliers on the page: Gemini 4 Argon Contributor 2x usage and Step 5 Preview 4x usage
    • Data: Gemini 4 Argon Contributor and the Muse Spark 1.2 and 1.3 Contributor models are marked Not ZDR; Contributor models may use prompts and completions for training in return for discounted pricing
    • The page states no single plan-wide dollar allowance. Its percentages apply to each model's monthly dollar limit; for a model with a $60 monthly limit (such as Kimi K3 on Go Plus) they equal $12 per 5 hours, $30 per week and $60 per month. Go Plus is $40/month on the same page.
    Verified 21 Oct 2026Source ↗Go pricing and FAQ ↗Buy on the official site
  • Zhipu (Z.ai)

    GLM Coding Plan Lite

    Coding planZ.ai (international)CreditsClear quotaOff-peak All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) Singapore Standard Time (UTC+8): Off-peak model usage is charged at 50% of the standard credit rate (page refers to model usage only, not MCP calls)
    $18
    per month
    or $14.40 per month, billed quarterly
    or $12.60 per month, billed yearly
    Limits
    2,000 credits per 5 hours10,000 credits per week
    Usage is metered in credits, not prompts: (input tokens x input multiplier + cached input tokens x cached multiplier + output tokens x output multiplier) / 10,000. GLM-5.3 multipliers are 6.9 input, 1.7 cached input and 24 output; GLM-5.3-Flash are 2.3, 0.56 and 8. Web Search, Web Reader and Zread MCP calls are charged as number of calls x 1.2. Models and MCP tools share one quota.
    Rules
    • May only be used within officially supported tools and products; use in unsupported tools may restrict benefits (docs.z.ai usage policy)
    • API calls made outside the supported tools do not draw on the plan quota (docs.z.ai overview)
    • Account sharing or multi-user access is prohibited (docs.z.ai usage policy)
    Models
    Works with
    • ZCode
    • Claude Code
    • Codex
    • OpenCode
    • AutoClaw
    • OpenClaw
    • Cline
    • Roo Code
    • Kilo Code
    • Crush
    • Goose
    • Cursor
    Worth knowing
    Off-peak: All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) (Singapore Standard Time (UTC+8)), Off-peak model usage is charged at 50% of the standard credit rate (page refers to model usage only, not MCP calls).
    More details
    • Peak hours: Monday to Friday, 14:00-18:00 Singapore Standard Time (UTC+8)
    • Upgrading resets credits to the new tier's limits immediately; renewing does not reset them (docs.bigmodel.cn overview)
    • Same plan as the BigModel (China) Lite entry, billed in USD for international customers. Requests for GLM-5.2 and GLM-5.1 are switched to GLM-5.3, and GLM-4.7 to GLM-5.3-Flash (docs overview). Team plans are not recorded.
    Note
    Verified 21 Oct 2026Source ↗Quick start (supported tools) ↗Usage policy ↗Plans, credits and multipliers ↗Buy on the official site
  • Xiaomi (MiMo)

    MiMo Token Plan Standard

    Coding planCreditsClear quotaOff-peak 00:00-08:00 UTC+8 (Beijing time; 16:00-24:00 UTC): Consumption coefficient 0.8x (20% fewer credits)
    ¥99
    per month
    or $16 per month
    or ¥1044 per year (¥87/mo)
    or $168.96 per year ($14.08/mo)
    Limits
    11,000,000,000 credits per month
    A fixed monthly quota of 11 billion credits, converted to tokens at the published per-model rates. When the quota runs out the service pauses; it does not draw on bonus or account balance. Unused quota does not carry over, and no 5-hour or weekly cap is listed.
    Rules
    • Quota may only be used in programming tools such as OpenClaw and OpenCode (Token Plan subscription page)
    • Prohibited: API calls for clearly non-coding scenarios, such as automated scripts and custom application backends; violations may lead to suspension of the subscription and blocking of the API key (Token Plan subscription page)
    Models
    Works with
    • Claude Code
    • OpenCode
    • OpenClaw
    Worth knowing
    Off-peak: 00:00-08:00 (UTC+8 (Beijing time; 16:00-24:00 UTC)), Consumption coefficient 0.8x (20% fewer credits).
    More details
    • First purchase: 12% off, once per account (Individual Edition)
    • Individual Edition only; the Team Edition (per seat) is not recorded. The plan page also lists Xiaomi's speech recognition (ASR) and text to speech (TTS) models.
    Verified 21 Oct 2026Source ↗Token Plan subscription terms ↗Buy on the official site
  • Meta

    High Usage

    Coding planMessagesOpaque: no published token or dollar conversion
    $15
    per month
    Limits
    High Usage is stated as 5x the Everyday Usage allowance, with more multimodal inputs. No separate prompt count is given for this tier. Subscriptions work only in the Muse Code CLI. the allowance is a multiple of the Everyday prompt range, with no published token or dollar conversion.
    Rules
    • Works only through the Muse Code CLI while signed in with your Meta Model API account
    • Benefits and availability vary by region; Muse Wonder and Muse Wonder Full are not available in South Korea and Muse Image is not available in India
    Models
    Works with
    • Muse Code CLI
    More details
    • More multimodal inputs
    • Subscriptions default to Contributor mode, which you can turn off at any time
    • Includes everything in Everyday Usage.
    Verified 21 Oct 2026Source ↗Muse Code subscriptions ↗Buy on the official site
  • Moonshot AI (Kimi)

    Plus (Kimi Code)

    Coding planCreditsOpaque credits: no published token conversion
    ¥99
    per month
    or ¥948 per year (¥79/mo)
    or $19 per month
    or $180 per year ($15/mo)
    Limits
    Plus and higher plans have a rolling 5-hour usage limit and no weekly limit, plus a monthly shared credit pool. The official pages give no numeric values. Once the monthly pool is used up, Kimi Code is frozen until it resets or the plan is upgraded. Unused credits do not roll over. Credits and usage limits with no published number or conversion to tokens or dollars.
    Rules
    • Quota is shared across all logged-in devices and API keys; devices inactive for over 30 days are unbound automatically (Kimi Code membership docs)
    • Paid services cannot be refunded or transferred once purchased and activated (Kimi help: Kimi Code benefits)
    Models
    Works with
    • Kimi Code CLI
    • Kimi Code Desktop
    • Kimi Code for VS Code
    • Claude Code
    • OpenCode
    • Codex
    • Hermes Agent
    • JetBrains (via ACP)
    • Zed (via ACP)
    More details
    • Parallel agent tasks: 2; projects: 20; scheduled/widget tasks: 10 (Kimi membership overview)
    • Extra Usage: optional paid balance for subscribed members; top-up minimum 25 CNY, up to 10 times and 3,000 CNY per day, balance cap 10,000 CNY (Kimi Code membership docs)
    • The Go tier (49 CNY per month) has no Kimi Code access and is not recorded here. The official membership.html page names Go, Plus and Pro as the new tiers; Max and Ultra appear on the membership overview page. The pricing page shows USD prices; the membership overview shows CNY.
    Verified 21 Oct 2026Source ↗Kimi Code membership ↗Kimi Code benefits ↗Membership pricing (USD) ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Standard

    Coding planInternational (Singapore region, USD)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models
    $18
    per month
    Limited-time price, regular $25
    or $195 per year ($16.25/mo)
    Limits
    45,000 credits per month
    A monthly quota of 45,000 credits; the weekly cap was removed on September 22, 2026. Credits per request depend on the model, token usage, thinking mode and tool calls, and the console shows per-call usage. 3-4 concurrent agents. Credits; Alibaba publishes no credit-to-token rate.
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan overview)
    • Available only in the Singapore region with Global deployment: inference runs globally and prompts and outputs cross borders (Token Plan overview)
    • For the subscriber's exclusive use; sharing the key with others may lead to suspension (Token Plan overview)
    • The Personal Edition does not support unsubscription; an expired subscription disables the API key (Token Plan overview)
    Models
    and 12 more
    • wan2.7-image-pro
    • wan3.0-video
    • deepseek-v4.1-flash
    • deepseek-v4.1-flash-1017
    • deepseek-v4-pro
    • deepseek-v4-pro-0813
    • deepseek-v4-flash-0731
    • glm-5.3
    • glm-5.2
    • happyhorse-1.1-i2v
    • happyhorse-1.1-t2v
    • happyhorse-1.1-r2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Concurrent agents: 1-2 (Lite), 2-3 (Essential), 3-4 (Standard), 6-8 (Pro)
    • Extra Bundle: $15 per bundle per month for 20,000 Credits; up to 5 held at once; needs an active subscription
    • Upgrades are prorated: you pay the price difference for the remaining days and get prorated extra credits
    • One subscription per verified identity per edition
    • The international edition of the Token Plan; the China edition (Beijing region, CNY) is a separate entry with the same credit quotas.
    Note
    Verified 21 Oct 2026Source ↗Token Plan landing page (prices) ↗Buy on the official site
  • MiniMax

    M Plan Go

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $22
    per month
    Limits
    Usage is measured relative to Go: Go is 1x (baseline), Explore is 3x and Build is 7.5x. Each tier has a 5-hour window and a weekly window; when a window ends, usage resets to the tier's full limit. Actual consumption varies with model, context length and task complexity. No token or dollar figures are published. Only relative multiples of Go are published; no conversion to tokens or dollars.
    Rules
    • Subscription key must be kept secret; do not commit it to a code repository (M Plan FAQ)
    • MiniMax may apply dynamic rate limiting and tighten limits during peak traffic; no peak hours or timezone are published (M Plan FAQ)
    Models
    Works with
    • MiniMax Code
    • Claude Code
    • Codex
    • Cursor
    • OpenCode
    • Pi
    • Hermes Agent
    • OpenClaw
    More details
    • Text model MiniMax-M3.1-Flash on all tiers; image and audio included on all tiers (M Plan page)
    • Video: Go does not include a video model; Explore and Build include H3 video, subject only to the weekly window (M Plan FAQ)
    • Annual billing: pay for 10 months and get 2 months free (M Plan FAQ); annual totals are not listed as amounts
    • Downgrades are not currently supported (M Plan FAQ)
    • Positioned by the FAQ for individual, interactive developer use; pay-as-you-go is recommended for production. The older Token Plan is closed to new purchases and is not recorded.
    Verified 21 Oct 2026Source ↗M Plan FAQ ↗M Plan overview ↗Token Plan notice (closed to new purchases) ↗Buy on the official site
  • Cursor

    Premium

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $30
    per month
    or $300 per year ($25/mo)
    Limits
    Paid plans include a weekly pool of usage for Cursor models, shared across Grok and Cursor. Premium is stated as '4x more usage compared to Plus'; no dollar or request amount is published. Third-party models are not in the pool: they are billed at the model's API list price from prepaid credit. Cursor publishes per-token rates for Cursor models beyond the pool. the pool is stated only as '4x Plus', with no dollar or request conversion; third-party usage is billed at published API list prices.
    Rules
    • Third-party models are billed separately from the plan price, at API list price from prepaid credit
    • Prices exclude tax. Cancel at least 24 hours before renewal to avoid additional charges
    Models
    and 46 more
    Works with
    • Cursor
    More details
    • Cloud agents and automations, code hosting and AI code reviews are included
    • When the weekly pool runs out you can wait for the reset, use prepaid credit, or upgrade; requests are not downgraded in quality or speed
    • The model list is the Cursor-pool models plus the third-party models the docs list for Premium and above; third-party names are as written on the models page, and some are marked hidden or Max Mode (legacy) there.
    Verified 21 Oct 2026Source ↗Pricing ↗Pricing and plans help ↗Buy on the official site
  • Moonshot AI (Kimi)

    Pro (Kimi Code)

    Coding planCreditsOpaque credits: no published token conversion
    ¥199
    per month
    or ¥1908 per year (¥159/mo)
    or $39 per month
    or $372 per year ($31/mo)
    Limits
    Plus and higher plans have a rolling 5-hour usage limit and no weekly limit, plus a monthly shared credit pool. The official pages give no numeric values. Once the monthly pool is used up, Kimi Code is frozen until it resets or the plan is upgraded. Unused credits do not roll over. Credits and usage limits with no published number or conversion to tokens or dollars.
    Rules
    • Quota is shared across all logged-in devices and API keys; devices inactive for over 30 days are unbound automatically (Kimi Code membership docs)
    • Paid services cannot be refunded or transferred once purchased and activated (Kimi help: Kimi Code benefits)
    Models
    • Kimi K3
    • Kimi K3 1M context
    • kimi-for-coding
    Works with
    • Kimi Code CLI
    • Kimi Code Desktop
    • Kimi Code for VS Code
    • Claude Code
    • OpenCode
    • Codex
    • Hermes Agent
    • JetBrains (via ACP)
    • Zed (via ACP)
    More details
    • K3 1M context and HighSpeed are included from Pro (Kimi Code membership docs)
    • Goal Mode and dedicated Kimi Claw are included in Pro and higher (Kimi help: Kimi Code benefits)
    • Parallel agent tasks: 2; projects: 20; scheduled/widget tasks: 15 (Kimi membership overview)
    • Extra Usage: optional paid balance for subscribed members; top-up minimum 25 CNY, up to 10 times and 3,000 CNY per day, balance cap 10,000 CNY
    • Official pages give no USD price for Pro.
    Verified 21 Oct 2026Source ↗Kimi Code membership ↗Kimi Code benefits ↗Membership pricing (USD) ↗Buy on the official site
  • OpenCode

    Go Plus

    Coding planDollar allowanceClear quota
    $40
    per month
    Limits
    20% of monthly limit per 5 hours50% of monthly limit per week100% of monthly limit per month
    Usage limits are defined as monthly dollar amounts, set per model. A model's 5-hour limit is 20% of its monthly limit, its weekly limit is 50% and its monthly limit is 100%. The page describes Go Plus as 'Higher limits and fewer interruptions'. The FAQ says these quotas are enforced across rolling windows.
    Rules
    • Usage limits may change as OpenCode learns from early usage and feedback
    • Cancel any time
    Models
    and 19 more
    • GLM-5.3
    • GLM-5.2
    • Kimi K3
    • Kimi K2.7 Code
    • LongCat-2.0
    • MiMo-V2.6-Pro
    • MiMo-V2.6-Flash
    • MiMo-V2.5-Pro
    • MiMo-V2.5
    • Qwen3.8 Max
    • Qwen3.8 Flash
    • Qwen3.7 Plus
    • MiniMax M3
    • Muse Spark 1.3 Contributor
    • Muse Spark 1.2 Contributor
    • DeepSeek V4.1 Flash
    • DeepSeek V4 Pro
    • Hy4 preview
    • Hy3
    Works with
    • OpenCode
    More details
    • Extra usage: turn it on in the Console Overview credits card; after the limits are reached, Go Plus falls back to the pay-as-you-go balance instead of blocking requests
    • Setup guide names Claude Code, Codex, Hermes, Pi and Raycast as agents to set up with the Go key
    • Refund possible if the charge was made within the last 14 days and the allowance is unused in that billing period
    • Data: Gemini 4 Argon Contributor and the Muse Spark 1.2 and 1.3 Contributor models are marked Not ZDR
    • The page gives one model list for Go and Go Plus (29 models per the FAQ) and states Go Plus only as higher limits; the Go Plus model list is therefore taken from the Go list and should be checked against the live page.
    Verified 21 Oct 2026Source ↗Go pricing and FAQ ↗Buy on the official site
  • Xiaomi (MiMo)

    MiMo Token Plan Pro

    Coding planCreditsClear quotaOff-peak 00:00-08:00 UTC+8 (Beijing time; 16:00-24:00 UTC): Consumption coefficient 0.8x (20% fewer credits)
    ¥329
    per month
    or $50 per month
    or ¥3468 per year (¥289/mo)
    or $528 per year ($44/mo)
    Limits
    38,000,000,000 credits per month
    A fixed monthly quota of 38 billion credits, converted to tokens at the published per-model rates. When the quota runs out the service pauses; it does not draw on bonus or account balance. Unused quota does not carry over, and no 5-hour or weekly cap is listed.
    Rules
    • Quota may only be used in programming tools such as OpenClaw and OpenCode (Token Plan subscription page)
    • Prohibited: API calls for clearly non-coding scenarios, such as automated scripts and custom application backends; violations may lead to suspension of the subscription and blocking of the API key (Token Plan subscription page)
    Models
    Works with
    • Claude Code
    • OpenCode
    • OpenClaw
    Worth knowing
    Off-peak: 00:00-08:00 (UTC+8 (Beijing time; 16:00-24:00 UTC)), Consumption coefficient 0.8x (20% fewer credits).
    More details
    • First purchase: 12% off, once per account (Individual Edition)
    • Individual Edition only; the Team Edition (per seat) is not recorded. The plan page also lists Xiaomi's speech recognition (ASR) and text to speech (TTS) models.
    Verified 21 Oct 2026Source ↗Token Plan subscription terms ↗Buy on the official site
  • Meta

    Power Usage

    Coding planMessagesOpaque: no published token or dollar conversion
    $50
    per month
    Limits
    Power Usage is stated as 20x the Everyday Usage allowance, with access to the most powerful Muse models, including Muse Wonder Full, early feature access and higher file uploads. No separate prompt count is given for this tier. Subscriptions work only in the Muse Code CLI. the allowance is a multiple of the Everyday prompt range, with no published token or dollar conversion.
    Rules
    • Works only through the Muse Code CLI while signed in with your Meta Model API account
    • Muse Wonder Full is available only on the Power Usage and Maximum Usage plans
    • Benefits and availability vary by region; Muse Wonder and Muse Wonder Full are not available in South Korea and Muse Image is not available in India
    Models
    Works with
    • Muse Code CLI
    More details
    • Early access to new features; higher file uploads
    • Subscriptions default to Contributor mode, which you can turn off at any time
    • Includes everything in High Usage.
    Verified 21 Oct 2026Source ↗Muse Code subscriptions ↗Muse Wonder ↗Buy on the official site
  • MiniMax

    M Plan Explore

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $55
    per month
    Limits
    Usage is measured relative to Go: Go is 1x, Explore is 3x and Build is 7.5x. Each tier has a 5-hour window and a weekly window; when a window ends, usage resets to the tier's full limit. Actual consumption varies with model, context length and task complexity. No token or dollar figures are published. Only relative multiples of Go are published; no conversion to tokens or dollars.
    Rules
    • Subscription key must be kept secret; do not commit it to a code repository (M Plan FAQ)
    • MiniMax may apply dynamic rate limiting and tighten limits during peak traffic; no peak hours or timezone are published (M Plan FAQ)
    Models
    Works with
    • MiniMax Code
    • Claude Code
    • Codex
    • Cursor
    • OpenCode
    • Pi
    • Hermes Agent
    • OpenClaw
    More details
    • Explore includes the H3 video model, subject only to the weekly window (M Plan FAQ)
    • Image and audio included (M Plan page); annual billing is 10 months paid for 12 (M Plan FAQ)
    • Upgrade paths: monthly subscribers may upgrade on either billing period; annual subscribers only on annual billing (M Plan FAQ)
    • Video is included in this subscription tier, so it is not API-only for M Plan. Pay-as-you-go video is also sold through the API (see the MiniMax API entry).
    Verified 21 Oct 2026Source ↗M Plan FAQ ↗Buy on the official site
  • Zhipu (Z.ai)

    GLM Coding Plan Pro

    Coding planZ.ai (international)CreditsClear quotaOff-peak All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) Singapore Standard Time (UTC+8): Off-peak model usage is charged at 50% of the standard credit rate
    $80
    per month
    or $64 per month, billed quarterly
    or $56 per month, billed yearly
    Limits
    12,000 credits per 5 hours60,000 credits per week
    Usage is metered in credits, not prompts: (input tokens x input multiplier + cached input tokens x cached multiplier + output tokens x output multiplier) / 10,000. GLM-5.3 multipliers are 6.9 input, 1.7 cached input and 24 output; GLM-5.3-Flash are 2.3, 0.56 and 8. Models and MCP tools share one quota.
    Rules
    • May only be used within officially supported tools and products (docs.z.ai usage policy)
    • API calls outside the supported tools do not draw on the plan quota (docs.z.ai overview)
    • Account sharing or multi-user access is prohibited (docs.z.ai usage policy)
    Models
    Works with
    • ZCode
    • Claude Code
    • Codex
    • OpenCode
    • AutoClaw
    • OpenClaw
    • Cline
    • Roo Code
    • Kilo Code
    • Crush
    • Goose
    • Cursor
    Worth knowing
    Off-peak: All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) (Singapore Standard Time (UTC+8)), Off-peak model usage is charged at 50% of the standard credit rate.
    More details
    • Concurrency limits are higher than Lite; page recommends Pro for 1-2 projects at a time (docs.z.ai usage policy)
    • Same plan as the BigModel (China) Pro entry, billed in USD for international customers. Requests for GLM-5.2 and GLM-5.1 are switched to GLM-5.3, and GLM-4.7 to GLM-5.3-Flash (docs overview). Team plans are not recorded.
    Note
    Verified 21 Oct 2026Source ↗Usage policy ↗Plans, credits and multipliers ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Pro

    Coding planInternational (Singapore region, USD)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models
    $68
    per month
    Limited-time price, regular $80
    or $770 per year ($64.17/mo)
    Limits
    180,000 credits per month
    A monthly quota of 180,000 credits; the weekly cap was removed on September 22, 2026. Credits per request depend on the model, token usage, thinking mode and tool calls, and the console shows per-call usage. 6-8 concurrent agents. Credits; Alibaba publishes no credit-to-token rate.
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan overview)
    • Available only in the Singapore region with Global deployment: inference runs globally and prompts and outputs cross borders (Token Plan overview)
    • For the subscriber's exclusive use; sharing the key with others may lead to suspension (Token Plan overview)
    • The Personal Edition does not support unsubscription; an expired subscription disables the API key (Token Plan overview)
    Models
    and 12 more
    • wan2.7-image-pro
    • wan3.0-video
    • deepseek-v4.1-flash
    • deepseek-v4.1-flash-1017
    • deepseek-v4-pro
    • deepseek-v4-pro-0813
    • deepseek-v4-flash-0731
    • glm-5.3
    • glm-5.2
    • happyhorse-1.1-i2v
    • happyhorse-1.1-t2v
    • happyhorse-1.1-r2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Concurrent agents: 1-2 (Lite), 2-3 (Essential), 3-4 (Standard), 6-8 (Pro)
    • Extra Bundle: $15 per bundle per month for 20,000 Credits; up to 5 held at once; needs an active subscription
    • Upgrades are prorated: you pay the price difference for the remaining days and get prorated extra credits
    • One subscription per verified identity per edition
    • The international edition of the Token Plan; the China edition (Beijing region, CNY) is a separate entry with the same credit quotas.
    Note
    Verified 21 Oct 2026Source ↗Token Plan landing page (prices) ↗Buy on the official site
  • Moonshot AI (Kimi)

    Max (Kimi Code)

    Coding planCreditsOpaque credits: no published token conversion
    ¥699
    per month
    or ¥6708 per year (¥559/mo)
    or $99 per month
    or $948 per year ($79/mo)
    Limits
    Plus and higher plans have a rolling 5-hour usage limit and no weekly limit, plus a monthly shared credit pool. The official pages give no numeric values. Once the monthly pool is used up, Kimi Code is frozen until it resets or the plan is upgraded. Unused credits do not roll over. Credits and usage limits with no published number or conversion to tokens or dollars.
    Rules
    • Quota is shared across all logged-in devices and API keys (Kimi Code membership docs)
    • Paid services cannot be refunded or transferred once purchased and activated (Kimi help: Kimi Code benefits)
    Models
    Works with
    • Kimi Code CLI
    • Kimi Code Desktop
    • Kimi Code for VS Code
    • Claude Code
    • OpenCode
    • Codex
    • Hermes Agent
    • JetBrains (via ACP)
    • Zed (via ACP)
    More details
    • Parallel agent tasks: 4; projects: 100; scheduled/widget tasks: 20 (Kimi membership overview)
    • Extra Usage: optional paid balance for subscribed members; top-up minimum 25 CNY, up to 10 times and 3,000 CNY per day, balance cap 10,000 CNY
    • Max appears on the kimi.com membership overview page with CNY prices. Its quota and model details beyond the Plus-and-higher rule were not separately listed on the pages read.
    Verified 21 Oct 2026Source ↗Kimi Code membership ↗Membership pricing (USD) ↗Buy on the official site
  • Cursor

    Super

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $100
    per month
    or $1000 per year ($83.33/mo)
    Limits
    Paid plans include a weekly pool of usage for Cursor models, shared across Grok and Cursor. Super is stated as '15x more usage compared to Plus' and adds priority model access at peak times over Premium; no dollar or request amount is published. Third-party models are billed separately at API list price from prepaid credit. the pool is stated only as '15x Plus', with no dollar or request conversion.
    Rules
    • Third-party models are billed separately from the plan price, at API list price from prepaid credit
    • Prices exclude tax. Cancel at least 24 hours before renewal to avoid additional charges
    Models
    and 46 more
    Works with
    • Cursor
    More details
    • Priority model access at peak times over Premium
    • When the weekly pool runs out you can wait for the reset, use prepaid credit, or upgrade; requests are not downgraded in quality or speed
    • Model list matches Premium (the Cursor docs list the same models for Premium and above; the pricing page adds 'priority model access at peak times' for Super).
    Verified 21 Oct 2026Source ↗Pricing ↗Pricing and plans help ↗Buy on the official site
  • Xiaomi (MiMo)

    MiMo Token Plan Max

    Coding planCreditsClear quotaOff-peak 00:00-08:00 UTC+8 (Beijing time; 16:00-24:00 UTC): Consumption coefficient 0.8x (20% fewer credits)
    ¥659
    per month
    or $100 per month
    or ¥6948 per year (¥579/mo)
    or $1056 per year ($88/mo)
    Limits
    82,000,000,000 credits per month
    A fixed monthly quota of 82 billion credits, converted to tokens at the published per-model rates. When the quota runs out the service pauses; it does not draw on bonus or account balance. Unused quota does not carry over, and no 5-hour or weekly cap is listed.
    Rules
    • Quota may only be used in programming tools such as OpenClaw and OpenCode (Token Plan subscription page)
    • Prohibited: API calls for clearly non-coding scenarios, such as automated scripts and custom application backends; violations may lead to suspension of the subscription and blocking of the API key (Token Plan subscription page)
    Models
    Works with
    • Claude Code
    • OpenCode
    • OpenClaw
    Worth knowing
    Off-peak: 00:00-08:00 (UTC+8 (Beijing time; 16:00-24:00 UTC)), Consumption coefficient 0.8x (20% fewer credits).
    More details
    • First purchase: 12% off, once per account (Individual Edition)
    • Individual Edition only; the Team Edition (per seat) is not recorded. The plan page also lists Xiaomi's speech recognition (ASR) and text to speech (TTS) models.
    Verified 21 Oct 2026Source ↗Token Plan subscription terms ↗Buy on the official site
  • Zhipu (Z.ai)

    GLM Coding Plan Max

    Coding planZ.ai (international)CreditsClear quotaOff-peak All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) Singapore Standard Time (UTC+8): Off-peak model usage is charged at 50% of the standard credit rate
    $168
    per month
    or $134.40 per month, billed quarterly
    or $117.60 per month, billed yearly
    Limits
    28,000 credits per 5 hours140,000 credits per week
    Usage is metered in credits, not prompts: (input tokens x input multiplier + cached input tokens x cached multiplier + output tokens x output multiplier) / 10,000. GLM-5.3 multipliers are 6.9 input, 1.7 cached input and 24 output; GLM-5.3-Flash are 2.3, 0.56 and 8. Models and MCP tools share one quota.
    Rules
    • May only be used within officially supported tools and products (docs.z.ai usage policy)
    • API calls outside the supported tools do not draw on the plan quota (docs.z.ai overview)
    • Account sharing or multi-user access is prohibited (docs.z.ai usage policy)
    Models
    Works with
    • ZCode
    • Claude Code
    • Codex
    • OpenCode
    • AutoClaw
    • OpenClaw
    • Cline
    • Roo Code
    • Kilo Code
    • Crush
    • Goose
    • Cursor
    Worth knowing
    Off-peak: All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) (Singapore Standard Time (UTC+8)), Off-peak model usage is charged at 50% of the standard credit rate.
    More details
    • Page recommends Max for developing 2 or more projects simultaneously (docs.z.ai usage policy)
    • Same plan as the BigModel (China) Max entry, billed in USD for international customers. Requests for GLM-5.2 and GLM-5.1 are switched to GLM-5.3, and GLM-4.7 to GLM-5.3-Flash (docs overview). Team plans are not recorded.
    Note
    Verified 21 Oct 2026Source ↗Usage policy ↗Plans, credits and multipliers ↗Buy on the official site
  • MiniMax

    M Plan Build

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $132
    per month
    Limits
    Usage is measured relative to Go: Go is 1x, Explore is 3x and Build is 7.5x. Each tier has a 5-hour window and a weekly window; when a window ends, usage resets to the tier's full limit. Actual consumption varies with model, context length and task complexity. No token or dollar figures are published. Only relative multiples of Go are published; no conversion to tokens or dollars.
    Rules
    • Subscription key must be kept secret; do not commit it to a code repository (M Plan FAQ)
    • MiniMax may apply dynamic rate limiting and tighten limits during peak traffic; no peak hours or timezone are published (M Plan FAQ)
    Models
    Works with
    • MiniMax Code
    • Claude Code
    • Codex
    • Cursor
    • OpenCode
    • Pi
    • Hermes Agent
    • OpenClaw
    More details
    • Build includes the H3 video model, subject only to the weekly window (M Plan FAQ)
    • Image and audio included (M Plan page); annual billing is 10 months paid for 12 (M Plan FAQ)
    • Video is included in this subscription tier, so it is not API-only for M Plan.
    Verified 21 Oct 2026Source ↗M Plan FAQ ↗Buy on the official site
  • Moonshot AI (Kimi)

    Ultra (Kimi Code)

    Coding planCreditsOpaque credits: no published token conversion
    ¥1399
    per month
    or ¥13428 per year (¥1119/mo)
    or $199 per month
    or $1908 per year ($159/mo)
    Limits
    Plus and higher plans have a rolling 5-hour usage limit and no weekly limit, plus a monthly shared credit pool. The official pages give no numeric values. Once the monthly pool is used up, Kimi Code is frozen until it resets or the plan is upgraded. Unused credits do not roll over. Credits and usage limits with no published number or conversion to tokens or dollars.
    Rules
    • Quota is shared across all logged-in devices and API keys (Kimi Code membership docs)
    • Paid services cannot be refunded or transferred once purchased and activated (Kimi help: Kimi Code benefits)
    Models
    Works with
    • Kimi Code CLI
    • Kimi Code Desktop
    • Kimi Code for VS Code
    • Claude Code
    • OpenCode
    • Codex
    • Hermes Agent
    • JetBrains (via ACP)
    • Zed (via ACP)
    More details
    • Parallel agent tasks: 4; projects: 100; scheduled/widget tasks: 25 (Kimi membership overview)
    • Extra Usage: optional paid balance for subscribed members; top-up minimum 25 CNY, up to 10 times and 3,000 CNY per day, balance cap 10,000 CNY
    • Ultra appears on the kimi.com membership overview page with CNY prices. Its quota and model details beyond the Plus-and-higher rule were not separately listed on the pages read.
    Verified 21 Oct 2026Source ↗Kimi Code membership ↗Membership pricing (USD) ↗Buy on the official site
  • Cursor

    Ultra

    Coding planUsage multiplesOpaque: no published token or dollar conversion
    $200
    per month
    or $2000 per year ($166.67/mo)
    Limits
    Paid plans include a weekly pool of usage for Cursor models, shared across Grok and Cursor. Ultra is stated as '50x more usage compared to Plus' and adds earliest access to all products and models over Super; no dollar or request amount is published. Third-party models are billed separately at API list price from prepaid credit. the pool is stated only as '50x Plus', with no dollar or request conversion.
    Rules
    • Third-party models are billed separately from the plan price, at API list price from prepaid credit
    • Prices exclude tax. Cancel at least 24 hours before renewal to avoid additional charges
    Models
    and 46 more
    Works with
    • Cursor
    More details
    • Earliest access to all products and models over Super
    • When the weekly pool runs out you can wait for the reset, use prepaid credit, or upgrade; requests are not downgraded in quality or speed
    • The legacy Ultra plan has the same name and $200/month price but is not sold to new customers; it is not recorded here.
    Verified 21 Oct 2026Source ↗Pricing ↗Pricing and plans help ↗Buy on the official site
  • Meta

    Maximum Usage

    Coding planMessagesOpaque: no published token or dollar conversion
    $200
    per month
    Limits
    Maximum Usage is stated as 80x the Everyday Usage allowance, with everything in Power Usage. No separate prompt count is given for this tier. At the limit there is no higher tier, so the subscriber must wait for the limit to refresh. Subscriptions work only in the Muse Code CLI. the allowance is a multiple of the Everyday prompt range, with no published token or dollar conversion.
    Rules
    • Works only through the Muse Code CLI while signed in with your Meta Model API account
    • On this plan you must wait for the limit to refresh once it is reached
    • Benefits and availability vary by region; Muse Wonder and Muse Wonder Full are not available in South Korea and Muse Image is not available in India
    Models
    Works with
    • Muse Code CLI
    More details
    • Subscriptions default to Contributor mode, which you can turn off at any time
    • Includes everything in Power Usage.
    Verified 21 Oct 2026Source ↗Muse Code subscriptions ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Lite

    Coding planChina (Beijing region, CNY)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash (page: 40% of normal cost); 50% off for deepseek-v4-pro-0813, deepseek-v4-flash-0731, deepseek-v4.1-flash and deepseek-v4.1-flash-1017
    ¥39
    per month
    Limited-time price, regular ¥60
    Limits
    11,500 credits per month
    A monthly quota of credits; the weekly cap was removed on September 22, 2026, and no 5-hour limit is listed. Credits per request depend on the model, token usage, thinking mode and tool calls, and the console shows per-call usage. Lite allows 1-2 concurrent agents. Credits; Alibaba publishes no credit-to-token rate (the page says credits are dynamically determined by model type, token usage, thinking mode and tool calls).
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan personal overview)
    • For the subscriber's exclusive use; sharing the key with others is prohibited (Token Plan personal overview)
    • Currently available only in the China (Beijing) region (Token Plan personal overview)
    • The Personal Edition does not support unsubscription; an expired subscription disables the API key (Token Plan personal overview)
    Models
    and 4 more
    • glm-5.2
    • qwen-image-3.0-pro
    • wan3.0-video
    • happyhorse-1.1-t2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash (page: 40% of normal cost); 50% off for deepseek-v4-pro-0813, deepseek-v4-flash-0731, deepseek-v4.1-flash and deepseek-v4.1-flash-1017.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Concurrent agents: 1-2 (Lite), 2-3 (Essential), 3-4 (Standard), 6-8 (Pro)
    • Extra Bundle: 100 CNY per bundle per month for 20,000 Credits; up to 5 held at once; needs an active subscription
    • Deduction order: plan quota first, then Extra Bundle Credits; when both are used up, service pauses until the next subscription month or an upgrade
    • Subscription is one per verified identity per edition
    • Names and prices come from the official help page (China site). The international edition (Singapore region, USD) is a separate entry.
    Verified 21 Oct 2026Source ↗Token Plan personal overview (English) ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Essential

    Coding planChina (Beijing region, CNY)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models
    ¥79
    per month
    Limited-time price, regular ¥120
    Limits
    25,500 credits per month
    A monthly quota of credits; the weekly cap was removed on September 22, 2026, and no 5-hour limit is listed. Credits per request depend on the model, token usage, thinking mode and tool calls. Essential allows 2-3 concurrent agents. Credits; Alibaba publishes no credit-to-token rate.
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan personal overview)
    • For the subscriber's exclusive use; sharing the key with others is prohibited (Token Plan personal overview)
    • Currently available only in the China (Beijing) region (Token Plan personal overview)
    Models
    and 4 more
    • glm-5.2
    • qwen-image-3.0-pro
    • wan3.0-video
    • happyhorse-1.1-t2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Extra Bundle: 100 CNY per bundle per month for 20,000 Credits; up to 5 held at once
    • Names and prices come from the official help page (China site). The international edition (Singapore region, USD) is a separate entry.
    Verified 21 Oct 2026Source ↗Token Plan personal overview (English) ↗Buy on the official site
  • Zhipu (BigModel)

    GLM Coding Plan Lite

    Coding planBigModel (China)CreditsClear quotaOff-peak All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) UTC+8 (Beijing time): Off-peak model calls are deducted at 50% of base credits
    ¥118
    per month
    or ¥94.40 per month, billed quarterly
    or ¥82.60 per month, billed yearly
    Limits
    2,000 credits per 5 hours10,000 credits per week
    Usage is metered in credits, not prompts: (input tokens x input multiplier + cached input tokens x cached multiplier + output tokens x output multiplier) / 10,000. GLM-5.3 multipliers are 6.9 input, 1.7 cached input and 24 output; GLM-5.3-Flash are 2.3, 0.56 and 8. Models and MCP tools share one quota.
    Rules
    • Only for use in officially designated tools and product environments; API calls in other tools do not draw on the plan quota (docs.bigmodel.cn overview)
    • For self-built apps, websites, bots or SaaS products, use the standard API, billed under its own terms (docs.bigmodel.cn FAQ)
    • The web Experience Center does not support the coding plan (docs.bigmodel.cn FAQ)
    Models
    Works with
    • ZCode
    • AutoClaw
    • Claude Code
    • Codex
    • OpenCode
    • WorkBuddy
    • Cherry Studio
    Worth knowing
    Off-peak: All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) (UTC+8 (Beijing time)), Off-peak model calls are deducted at 50% of base credits.
    More details
    • OpenClaw is supported on secondary scheduling with best-effort delivery; Coding Agent tasks get priority (docs.bigmodel.cn overview)
    • Same plan as the Z.ai (international) Lite entry, billed in CNY for China. Requests for GLM-5.2 and GLM-5.1 are switched to GLM-5.3, and GLM-5-Turbo and GLM-4.7 to GLM-5.3-Flash (docs overview). The FAQ requires the standard API for self-built apps, websites, bots and SaaS. Team plans are not recorded.
    Note
    Verified 21 Oct 2026Source ↗FAQ ↗套餐概览 (plans, credits and multipliers) ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Standard

    Coding planChina (Beijing region, CNY)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models
    ¥139
    per month
    Limited-time price, regular ¥180
    Limits
    45,000 credits per month
    A monthly quota of credits; the weekly cap was removed on September 22, 2026, and no 5-hour limit is listed. Credits per request depend on the model, token usage, thinking mode and tool calls. Standard allows 3-4 concurrent agents and includes Harness benefits (metered separately). Credits; Alibaba publishes no credit-to-token rate.
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan personal overview)
    • For the subscriber's exclusive use; sharing the key with others is prohibited (Token Plan personal overview)
    • Currently available only in the China (Beijing) region (Token Plan personal overview)
    Models
    and 4 more
    • glm-5.2
    • qwen-image-3.0-pro
    • wan3.0-video
    • happyhorse-1.1-t2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Extra Bundle: 100 CNY per bundle per month for 20,000 Credits; up to 5 held at once
    • Names and prices come from the official help page (China site). The international edition (Singapore region, USD) is a separate entry.
    Verified 21 Oct 2026Source ↗Token Plan personal overview (English) ↗Buy on the official site
  • Zhipu (BigModel)

    GLM Coding Plan Pro

    Coding planBigModel (China)CreditsClear quotaOff-peak All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) UTC+8 (Beijing time): Off-peak model calls are deducted at 50% of base credits
    ¥538
    per month
    or ¥430.40 per month, billed quarterly
    or ¥376.60 per month, billed yearly
    Limits
    12,000 credits per 5 hours60,000 credits per week
    Usage is metered in credits, not prompts: (input tokens x input multiplier + cached input tokens x cached multiplier + output tokens x output multiplier) / 10,000. GLM-5.3 multipliers are 6.9 input, 1.7 cached input and 24 output; GLM-5.3-Flash are 2.3, 0.56 and 8. Models and MCP tools share one quota.
    Rules
    • Only for use in officially designated tools and product environments (docs.bigmodel.cn overview)
    • For self-built apps, websites, bots or SaaS products, use the standard API, billed under its own terms (docs.bigmodel.cn FAQ)
    Models
    Works with
    • ZCode
    • AutoClaw
    • Claude Code
    • Codex
    • OpenCode
    • WorkBuddy
    • Cherry Studio
    Worth knowing
    Off-peak: All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) (UTC+8 (Beijing time)), Off-peak model calls are deducted at 50% of base credits.
    More details
    • Same plan as the Z.ai (international) Pro entry, billed in CNY for China. Requests for GLM-5.2 and GLM-5.1 are switched to GLM-5.3, and GLM-5-Turbo and GLM-4.7 to GLM-5.3-Flash (docs overview). The FAQ requires the standard API for self-built apps, websites, bots and SaaS. Team plans are not recorded.
    Note
    Verified 21 Oct 2026Source ↗FAQ ↗套餐概览 (plans, credits and multipliers) ↗Buy on the official site
  • Alibaba Cloud (Model Studio)

    Token Plan (Personal) Pro

    Coding planChina (Beijing region, CNY)CreditsOpaque credits: no published token conversionOff-peak 22:00-08:00 (every night, crossing midnight) UTC+8: 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models
    ¥499
    per month
    Limited-time price, regular ¥600
    Limits
    180,000 credits per month
    A monthly quota of credits; the weekly cap was removed on September 22, 2026, and no 5-hour limit is listed. Credits per request depend on the model, token usage, thinking mode and tool calls. Pro allows 6-8 concurrent agents. Credits; Alibaba publishes no credit-to-token rate.
    Rules
    • Interactive use only inside coding and agent tools; must not be used for automation scripts, custom application backends or any non-interactive batch calls (Token Plan personal overview)
    • For the subscriber's exclusive use; sharing the key with others is prohibited (Token Plan personal overview)
    • Currently available only in the China (Beijing) region (Token Plan personal overview)
    Models
    and 4 more
    • glm-5.2
    • qwen-image-3.0-pro
    • wan3.0-video
    • happyhorse-1.1-t2v
    Works with
    • Claude Code
    • Cursor
    • Qwen Code
    • Qoder
    • Qoder CN
    • OpenClaw
    Worth knowing
    Off-peak: 22:00-08:00 (every night, crossing midnight) (UTC+8), 60% off Credits for qwen3.8-max and qwen3.8-flash; 50% off for the listed deepseek-v4 models.
    Rollover: Unused Credits do not carry over to the next cycle
    More details
    • Extra Bundle: 100 CNY per bundle per month for 20,000 Credits; up to 5 held at once
    • Names and prices come from the official help page (China site). The international edition (Singapore region, USD) is a separate entry.
    Verified 21 Oct 2026Source ↗Token Plan personal overview (English) ↗Buy on the official site
  • Zhipu (BigModel)

    GLM Coding Plan Max

    Coding planBigModel (China)CreditsClear quotaOff-peak All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) UTC+8 (Beijing time): Off-peak model calls are deducted at 50% of base credits
    ¥1078
    per month
    or ¥862.40 per month, billed quarterly
    or ¥754.60 per month, billed yearly
    Limits
    28,000 credits per 5 hours140,000 credits per week
    Usage is metered in credits, not prompts: (input tokens x input multiplier + cached input tokens x cached multiplier + output tokens x output multiplier) / 10,000. GLM-5.3 multipliers are 6.9 input, 1.7 cached input and 24 output; GLM-5.3-Flash are 2.3, 0.56 and 8. Models and MCP tools share one quota.
    Rules
    • Only for use in officially designated tools and product environments (docs.bigmodel.cn overview)
    • For self-built apps, websites, bots or SaaS products, use the standard API, billed under its own terms (docs.bigmodel.cn FAQ)
    Models
    Works with
    • ZCode
    • AutoClaw
    • Claude Code
    • Codex
    • OpenCode
    • WorkBuddy
    • Cherry Studio
    Worth knowing
    Off-peak: All hours outside Monday to Friday 14:00-18:00 (peak hours are charged 1x) (UTC+8 (Beijing time)), Off-peak model calls are deducted at 50% of base credits.
    More details
    • Same plan as the Z.ai (international) Max entry, billed in CNY for China. Requests for GLM-5.2 and GLM-5.1 are switched to GLM-5.3, and GLM-5-Turbo and GLM-4.7 to GLM-5.3-Flash (docs overview). The FAQ requires the standard API for self-built apps, websites, bots and SaaS. Team plans are not recorded.
    Note
    Verified 21 Oct 2026Source ↗FAQ ↗套餐概览 (plans, credits and multipliers) ↗Buy on the official site

Chat & general

17 entries, cheapest first

  • xAI

    Plus (The X Subscription)

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $8
    per month
    or $80 per year ($6.67/mo)
    Limits
    Each paid seat on The X Subscription gets a weekly pool for eligible first-party usage across supported Grok and Cursor surfaces. Plus is the base level. xAI says usage rates vary by product, and publishes no numeric pool size on the pages read. The pool refreshes weekly, and top-ups are available once it runs out. Weekly pool with no published size or token or dollar conversion.
    Rules
    • Grok Bot is not included on Plus (it requires Premium or higher)
    • Third-party models in Cursor are not part of the weekly pool and are paid as Additional Usage
    Works with
    • Grok apps
    • Grok tab on X
    • Cursor
    More details
    • Checkmark, creator tools and fewer ads on X (per x.ai announcement)
    • Top-ups buy Additional Usage credits for usage beyond what is included
    • The X Subscription launched Oct 12, 2026 as one bundle of Grok, Cursor and X. New customers start on it, and existing SuperGrok, Cursor and X Premium plans continue unless the user changes them. xAI's official announcement names no specific models. x.ai/pricing lists the same prices. SuperGrok and SuperGrok Heavy are not sold on x.ai/pricing or grok.com/plans.
    Verified 21 Oct 2026Source ↗Introducing The X Subscription (Oct 12, 2026) ↗Usage limits ↗Grok Bot ↗x.ai pricing ↗grok.com plans ↗Buy on the official site
  • OpenAI

    ChatGPT Go

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $8
    per month
    Limits
    Go includes Codex for lightweight coding tasks, with GPT-6 Luna at Standard speed in the desktop app, subject to rollout. The pricing docs give no numeric Codex limit for Go. No Codex allowance is published for Go, and the pricing docs give no conversion to tokens or dollars.
    Models
    • GPT-6 Luna
    Works with
    • Codex desktop app (GPT-6 Luna, subject to rollout)
    More details
    • Included here because OpenAI's pricing docs list Go as a ChatGPT tier with Codex. Free is also a ChatGPT tier with Codex but is not a purchase, so it has no entry.
    Verified 21 Oct 2026Source ↗Buy on the official site
  • Anthropic

    Pro

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $20
    per month
    or $200 per year ($16.67/mo)
    Limits
    Usage limits reset on a rolling five-hour session window, and paid plans add a weekly limit on top. Anthropic says Pro gives at least 5x more usage per 5-hour session than Free, and that there is no fixed message count. Claude web, desktop, mobile and Claude Code draw from the same limits. When a limit is reached, usage credits can keep work going at standard API rates. Usage is described as multiples of Free and as session and weekly windows; Anthropic publishes no token, message or dollar amount for Pro limits.
    Models
    Works with
    • Claude Code (terminal and supported IDEs)
    • Claude web, desktop and mobile apps
    More details
    • Claude Code included in all paid plans
    • Anthropic's May 6, 2026 post says Claude Code five-hour rate limits were doubled for Pro and Max and the peak-hours limit reduction was removed for those accounts
    • The pricing page's comparison table shows Fable 5.1 on Pro only through usage credits. Claude Code limits are shared with the Claude apps.
    Verified 21 Oct 2026Source ↗Anthropic news: higher usage limits (May 6, 2026) ↗Use Claude Code with your Pro or Max plan ↗Buy on the official site
  • Google

    Google AI Pro

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $19.99
    per month
    Limits
    Google says AI Pro has 4x higher usage access than Free. Gemini app limits are compute-based: they refresh every 5 hours until a weekly limit is reached, and depend on prompt complexity, models and features used, and chat length. Google says limits may change without notice, and that usage can be extended by buying AI credits. Google publishes multiples and compute-based limits, not request or token counts.
    Rules
    • Gemini CLI and the Gemini Code Assist IDE extensions stopped serving Google AI Pro consumer accounts on June 18, 2026. Sign-in with Google no longer works for them (Google deprecation page). Google points users to Antigravity.
    Models
    Works with
    • Gemini app
    • Google Antigravity
    More details
    • 5 TB storage (Google AI Plans page)
    • $10 monthly Google Cloud credits from the Google Developer Program (Google AI Plans page)
    • Expanded rate limits to the agent model in Google Antigravity
    • The one.google.com plans page loaded without prices in the fetch, so the $19.99 figure comes from gemini.google/subscriptions. Gemini 3.1 Pro is named on the Google AI Plans page. The subscriptions page names the Pro and Flash models only generically. No Gemini CLI request limit for this tier is stated on a current official page (see unverified).
    Verified 21 Oct 2026Source ↗Gemini Code Assist for individuals deprecation ↗Google AI plans ↗Antigravity pricing ↗Buy on the official site
  • OpenAI

    ChatGPT Plus

    Chat & generalMessagesOpaque: no published token or dollar conversion
    $20
    per month
    Limits
    Local messages and cloud chats share the plan's usage allowance. OpenAI gives estimated local-message ranges per five-hour window that vary by model, and says these are estimates, not fixed message limits. Weekly limits may also apply. The usage dashboard shows current values, and the Codex CLI /status command shows them during a session. Estimated message ranges per five-hour window with no published token or dollar conversion for included usage.
    Models
    Works with
    • Codex web
    • Codex CLI
    • Codex IDE extension
    • Codex iOS app
    More details
    • Cloud integrations for Codex
    • Can buy additional credits when limits are reached, without upgrading
    • Codex access comes with the ChatGPT plan; OpenAI sells no separate Codex-only subscription. Plus includes limited GPT-6 Astra usage, with credits for more (Codex pricing FAQ). The per-5-hour estimates also cover Standard Business.
    Verified 21 Oct 2026Source ↗Codex pricing ↗Buy on the official site
  • Meta

    Power plan

    Chat & generalTokensOpaque: no published token or dollar conversion
    $20
    per month
    Limits
    500,000,000 Muse tokens per week
    Muse is free with a usage limit. The weekly allowance for Muse increases as you upgrade plans; the Power plan includes 500M Muse tokens per week. The limit refreshes over time. the allowance is in 'Muse tokens'; the page does not define these against API tokens or dollars.
    Rules
    • Muse and Muse subscriptions are in limited testing and are not available in all locations yet (meta.com help)
    • Benefits and availability can vary by region and account (meta.com help)
    Works with
    • Muse (mobile app and muse.ai)
    More details
    • This is Meta's consumer Muse personal-agent subscription, separate from Muse Code subscriptions and the Meta Model API. The help page lists no models and no features beyond the weekly allowance.
    Verified 21 Oct 2026Source ↗Buy on the official site
  • xAI

    Premium (The X Subscription)

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $30
    per month
    or $300 per year ($25/mo)
    Limits
    Premium gives a weekly pool of 4x the Plus plan's usage, covering eligible first-party Grok and Cursor usage. The Grok Bot usage draws from the same pool. xAI publishes no numeric pool size on the pages read. The pool refreshes weekly, and top-ups are available once it runs out. Usage multiple vs Plus with no published pool size or token or dollar conversion.
    Rules
    • Third-party models in Cursor are not part of the weekly pool and are paid as Additional Usage
    Works with
    • Grok apps
    • Grok tab on X
    • Cursor
    • Grok Bot
    More details
    • Benefits of SuperGrok, Cursor Pro and X Premium+ in one plan, previously $90/month combined (x.ai announcement)
    • Grok Bot included
    • Top-ups buy Additional Usage credits for usage beyond what is included
    • Premium is the lowest tier with Grok Bot. Team seats are $30 per seat on Premium. The x.ai announcement says the Premium benefits of SuperGrok, Cursor Pro and X Premium+ were previously $90/month combined. xAI's official announcement names no specific models. grok.com/plans showed a first-month offer of 50% off on 2026-10-21.
    Verified 21 Oct 2026Source ↗Introducing The X Subscription (Oct 12, 2026) ↗Usage limits ↗Grok Bot ↗x.ai pricing ↗grok.com plans ↗Buy on the official site
  • xAI

    Super (The X Subscription)

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $100
    per month
    or $1000 per year ($83.33/mo)
    Limits
    Super gives a weekly pool of 15x the Plus plan's usage, covering eligible first-party Grok and Cursor usage, and includes all Premium features. xAI says higher tiers include larger pools and publishes no numeric pool size on the pages read. The pool refreshes weekly, and top-ups are available once it runs out. Usage multiple vs Plus with no published pool size or token or dollar conversion.
    Rules
    • Third-party models in Cursor are not part of the weekly pool and are paid as Additional Usage
    Works with
    • Grok apps
    • Grok tab on X
    • Cursor
    • Grok Bot
    More details
    • All Premium features
    • Grok Bot included
    • Team seats listed at $100 per seat with the same usage as Super
    • The x.ai announcement describes Super as '15x Plus Plan' with 'All of the above features'. xAI's official announcement names no specific models. grok.com/plans showed a first-month offer of 50% off on 2026-10-21.
    Verified 21 Oct 2026Source ↗Introducing The X Subscription (Oct 12, 2026) ↗Usage limits ↗x.ai pricing ↗grok.com plans ↗Buy on the official site
  • Google

    Google AI Ultra 5x

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $99.99
    per month
    Limits
    Ultra 5x gives 5x higher usage limits than AI Pro. Google's help page says Ultra gets 5x or 20x the Pro quota in Gemini and Antigravity, depending on the subscription. Gemini app limits refresh every 5 hours until a weekly limit is reached, and depend on prompt complexity, models and features used, and chat length. Limits may change without notice. Google publishes multiples relative to AI Pro, not request or token counts.
    Rules
    • Gemini CLI and the Gemini Code Assist IDE extensions stopped serving Google AI Ultra consumer accounts on June 18, 2026. Sign-in with Google no longer works for them (Google deprecation page). Google points users to Antigravity.
    Models
    Works with
    • Gemini app
    • Google Antigravity
    More details
    • Higher access to Gemini 4 Argon in Google Antigravity
    • Storage: 'Starting at 20 TB' (Google AI Plans page; the page's tier table lists 20 TB and 30 TB across the two Ultra tiers, mapping not confirmed)
    • $40 monthly Google Cloud credits from the Google Developer Program are listed under Google AI Ultra (the page's table also shows $100 monthly credits; tier mapping not confirmed)
    • Google lists this tier as one of two Ultra prices. The price on gemini.google/subscriptions is listed as 'Starting at'. The one.google.com plans page loaded without prices in the fetch.
    Verified 21 Oct 2026Source ↗Gemini Code Assist for individuals deprecation ↗Google AI plans ↗Antigravity pricing ↗Buy on the official site
  • OpenAI

    ChatGPT Pro 100

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $100
    per month
    Limits
    Pro plans currently have no five-hour limit according to OpenAI's pricing docs. Weekly limits may also apply, and the included Codex allowance is not published as a number. OpenAI points users to the usage dashboard for current limits. No five-hour window and no published Codex allowance for this tier, so there is no conversion to tokens or dollars.
    Rules
    • Sharing account credentials, reselling access and automatically extracting data are prohibited (help center, About ChatGPT Pro tiers)
    Models
    Works with
    • Codex web
    • Codex CLI
    • Codex IDE extension
    • Codex iOS app
    More details
    • Everything in Plus
    • Fast mode: 2.5x included subscription usage (OpenAI pricing docs)
    • Can buy additional credits when limits are reached
    • OpenAI names the tiers Pro 100, Pro 200 and Pro 500. It says Pro 200 includes more usage than Pro 100 and Pro 500 the most, and publishes no amounts. GPT-6 Astra is included in the Codex allowance on the $100 and $200 Pro plans (Codex pricing FAQ).
    Verified 21 Oct 2026Source ↗About ChatGPT Pro tiers ↗Codex pricing ↗Buy on the official site
  • Anthropic

    Max 5x

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $100
    per month
    Limits
    Max 5x gives five times the Pro plan's per-session usage allowance, over a rolling five-hour session window. It also has a weekly limit across all models and a separate weekly limit for Fable. Anthropic says there is no fixed message count, and that it may add other caps at its discretion. Usage credits, billed at standard API rates, can continue work after the included limits run out. Anthropic states the limit as a multiple of Pro's per-session allowance and publishes no token, message or dollar amount.
    Rules
    • Monthly API credits ($100 a month on Max 5x) are for building on the Claude Platform and cannot be used in Claude Code or the Claude apps, including for extra usage (Anthropic support article)
    • Anthropic may impose weekly or monthly caps or model and feature limits at its discretion (Max plan support article)
    Models
    Works with
    • Claude Code (terminal and supported IDEs)
    • Claude web, desktop and mobile apps
    More details
    • Higher output limits for all tasks
    • Priority access at high traffic times
    • Early access to advanced Claude features
    • Monthly API credits of $100 for the Claude Platform (per the pricing page and Max plan support article)
    • Anthropic's May 6, 2026 post says Claude Code five-hour rate limits were doubled for Pro and Max and the peak-hours limit reduction was removed
    • The Max plan support article gives the $100 price for Max 5x. The pricing page shows only 'From $100' and '$100 to $200/month in API credits'. Fable 5.1 appears in the pricing page's Max cells as '50% of weekly limits*'. The support article does not mention Fable.
    Verified 21 Oct 2026Source ↗Plans & pricing ↗Use Claude Code with your Pro or Max plan ↗Anthropic news: higher usage limits (May 6, 2026) ↗Buy on the official site
  • Meta

    Maximum plan

    Chat & generalTokensOpaque: no published token or dollar conversion
    $100
    per month
    Limits
    3,000,000,000 Muse tokens per week
    Muse is free with a usage limit. The weekly allowance for Muse increases as you upgrade plans; the Maximum plan includes 3B Muse tokens per week. The limit refreshes over time. the allowance is in 'Muse tokens'; the page does not define these against API tokens or dollars.
    Rules
    • Muse and Muse subscriptions are in limited testing and are not available in all locations yet (meta.com help)
    • Benefits and availability can vary by region and account (meta.com help)
    Works with
    • Muse (mobile app and muse.ai)
    More details
    • Consumer Muse personal-agent subscription, separate from Muse Code subscriptions and the Meta Model API. The help page lists no models.
    Verified 21 Oct 2026Source ↗Buy on the official site
  • xAI

    Ultra (The X Subscription)

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $200
    per month
    or $2000 per year ($166.67/mo)
    Limits
    Ultra gives a weekly pool of 50x the Plus plan's usage, covering eligible first-party Grok and Cursor usage, and includes all Premium and Super features. xAI says higher tiers include larger pools and publishes no numeric pool size on the pages read. The pool refreshes weekly, and top-ups are available once it runs out. Usage multiple vs Plus with no published pool size or token or dollar conversion.
    Rules
    • Third-party models in Cursor are not part of the weekly pool and are paid as Additional Usage
    Works with
    • Grok apps
    • Grok tab on X
    • Cursor
    • Grok Bot
    More details
    • All Premium and Super features
    • Grok Bot included
    • Team seats listed at $200 per seat with the same usage as Ultra
    • X ad removal from the timeline on higher tiers (x.ai announcement: 'higher tiers removing ads from the timeline entirely')
    • The x.ai announcement describes Ultra as '50x Plus Plan' with 'All of the above features'. xAI's official announcement names no specific models. grok.com/plans showed a first-month offer of 50% off on 2026-10-21.
    Verified 21 Oct 2026Source ↗Introducing The X Subscription (Oct 12, 2026) ↗Usage limits ↗x.ai pricing ↗grok.com plans ↗Buy on the official site
  • Google

    Google AI Ultra 20x

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $199.99
    per month
    Limits
    Ultra 20x gives 20x higher usage limits than AI Pro. Google's help page says Ultra gets 5x or 20x the Pro quota in Gemini and Antigravity, depending on the subscription. Gemini app limits refresh every 5 hours until a weekly limit is reached, and depend on prompt complexity, models and features used, and chat length. Limits may change without notice. Google publishes multiples relative to AI Pro, not request or token counts.
    Rules
    • Gemini CLI and the Gemini Code Assist IDE extensions stopped serving Google AI Ultra consumer accounts on June 18, 2026. Sign-in with Google no longer works for them (Google deprecation page). Google points users to Antigravity.
    Models
    Works with
    • Gemini app
    • Google Antigravity
    More details
    • Highest access to Gemini 4 Argon in Google Antigravity, with the highest limits and prioritized traffic (Google AI Ultra help page)
    • Storage: 30 TB is listed in the tier table for one Ultra tier (Google AI Plans page; mapping not confirmed)
    • $100 monthly Google Cloud credits appear in the page's table (tier mapping not confirmed)
    • Same Ultra family as Ultra 5x, at the 20x multiple.
    Verified 21 Oct 2026Source ↗Gemini Code Assist for individuals deprecation ↗Google AI plans ↗Antigravity pricing ↗Buy on the official site
  • OpenAI

    ChatGPT Pro 200

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $200
    per month
    Limits
    Pro plans currently have no five-hour limit according to OpenAI's pricing docs. Weekly limits may also apply, and the included Codex allowance is not published as a number. OpenAI points users to the usage dashboard for current limits. No five-hour window and no published Codex allowance for this tier, so there is no conversion to tokens or dollars.
    Rules
    • Sharing account credentials, reselling access and automatically extracting data are prohibited (help center, About ChatGPT Pro tiers)
    Models
    Works with
    • Codex web
    • Codex CLI
    • Codex IDE extension
    • Codex iOS app
    More details
    • Everything in Plus
    • Fast mode: 2.5x included subscription usage (OpenAI pricing docs)
    • Can buy additional credits when limits are reached
    • OpenAI names the tiers Pro 100, Pro 200 and Pro 500. It says Pro 200 includes more usage than Pro 100 and Pro 500 the most, and publishes no amounts. GPT-6 Astra is included in the Codex allowance on the $100 and $200 Pro plans (Codex pricing FAQ). New Pro 200 subscriptions get a lower allowance than before; subscribers active between September 22 and September 29, 2026 keep the previous allowance through October 29, 2026.
    Verified 21 Oct 2026Source ↗About ChatGPT Pro tiers ↗Codex pricing ↗Buy on the official site
  • Anthropic

    Max 20x

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $200
    per month
    Limits
    Max 20x gives twenty times the Pro plan's per-session usage allowance, over a rolling five-hour session window. It also has a weekly limit across all models and a separate weekly limit for Fable. Anthropic says there is no fixed message count, and that it may add other caps at its discretion. Usage credits, billed at standard API rates, can continue work after the included limits run out. Anthropic states the limit as a multiple of Pro's per-session allowance and publishes no token, message or dollar amount.
    Rules
    • Monthly API credits ($200 a month on Max 20x) are for building on the Claude Platform and cannot be used in Claude Code or the Claude apps, including for extra usage (Anthropic support article)
    • Anthropic may impose weekly or monthly caps or model and feature limits at its discretion (Max plan support article)
    Models
    Works with
    • Claude Code (terminal and supported IDEs)
    • Claude web, desktop and mobile apps
    More details
    • Higher output limits for all tasks
    • Priority access at high traffic times
    • Early access to advanced Claude features
    • Monthly API credits of $200 for the Claude Platform (per the pricing page and Max plan support article)
    • Anthropic's May 6, 2026 post says Claude Code five-hour rate limits were doubled for Pro and Max and the peak-hours limit reduction was removed
    • Same plan family as Max 5x, with the 20x multiple. The Max plan support article gives the $200 price. The pricing page shows only 'From $100' for Max.
    Verified 21 Oct 2026Source ↗Plans & pricing ↗Use Claude Code with your Pro or Max plan ↗Anthropic news: higher usage limits (May 6, 2026) ↗Buy on the official site
  • OpenAI

    ChatGPT Pro 500

    Chat & generalUsage multiplesOpaque: no published token or dollar conversion
    $500
    per month
    Limits
    Pro plans currently have no five-hour limit according to OpenAI's pricing docs. Weekly limits may also apply, and the included Codex allowance is not published as a number. OpenAI points users to the usage dashboard for current limits. No five-hour window and no published Codex allowance for this tier, so there is no conversion to tokens or dollars.
    Rules
    • Sharing account credentials, reselling access and automatically extracting data are prohibited (help center, About ChatGPT Pro tiers)
    Models
    Works with
    • Codex web
    • Codex CLI
    • Codex IDE extension
    • Codex iOS app
    More details
    • Everything in Plus
    • Ultrafast mode on this tier: 8x included subscription usage for GPT-6.1 Sol and GPT-6 Astra (OpenAI pricing docs)
    • Fast mode: 2.5x included subscription usage
    • Can buy additional credits when limits are reached
    • OpenAI names the tiers Pro 100, Pro 200 and Pro 500. It says Pro 200 includes more usage than Pro 100 and Pro 500 the most, and publishes no amounts. The only Pro plan with Ultrafast, for GPT-6 Astra and GPT-6.1 Sol: it uses the included allowance at 8x the Standard rate, then purchased credits at 6x.
    Verified 21 Oct 2026Source ↗About ChatGPT Pro tiers ↗Codex pricing ↗Buy on the official site

Bundle / gateway

2 entries, cheapest first

  • OpenCode

    Zen

    Bundle / gatewayPay per tokenClear quota
    Pay per use
    Limits
    Pay-as-you-go: you add a prepaid balance and are charged per request at the published per-1M-token rates for each model. The page says you can set monthly spend limits in your account. Free models are listed at 0 for input, output and cached read.
    API rates
    ModelInputOutput
    Jev 1.13$0.042$0per 1M tokens, Output listed as Free; no cached-read rate listed
    MiniMax M3$0.30$1.20per 1M tokens, Cached read $0.06
    MiniMax M2.7$0.30$1.20per 1M tokens, Cached read $0.06
    MiniMax M2.5$0.30$1.20per 1M tokens, Cached read $0.06
    GLM 5.3 Flash$0.15$0.50per 1M tokens, Cached read $0.03
    GLM 5.3$1.40$4.40per 1M tokens, Cached read $0.26
    GLM 5.2$1.40$4.40per 1M tokens, Cached read $0.26
    GLM 5.1$1.40$4.40per 1M tokens, Cached read $0.26
    and 95 more rates
    GLM 5$1$3.20per 1M tokens, Cached read $0.20
    Kimi K2.7 Code$0.95$4per 1M tokens, Cached read $0.19
    Kimi K3$3$15per 1M tokens, Cached read $0.30
    Kimi K2.6$0.95$4per 1M tokens, Cached read $0.16
    Kimi K2.5$0.60$3per 1M tokens, Cached read $0.10
    Mistral Large 4$0.68$2.09per 1M tokens, 50% off (was $1.36 input, $4.18 output); cached read $0.07 (was $0.14)
    Qwen3.8 Max$2$6per 1M tokens, Cached read $0.25; cache write $2.50
    Qwen3.8 Flash$0.15$0.47per 1M tokens, Cached read $0.016; cache write $0.20
    Qwen3.7 Max$2.50$7.50per 1M tokens, Cached read $0.50; cache write $3.125
    Qwen3.7 Plus$0.40$1.60per 1M tokens, Cached read $0.04; cache write $0.50
    Qwen3.6 Plus$0.50$3per 1M tokens, Cached read $0.05; cache write $0.625
    Qwen3.5 Plus$0.20$1.20per 1M tokens, Cached read $0.02; cache write $0.25
    DeepSeek V4.1 Flash$0.30$1.20per 1M tokens, Cached read $0.006
    DeepSeek V4 Pro$1.74$3.48per 1M tokens, Cached read $0.145
    Claude Fable 5.1$10$50per 1M tokens, Cached read $0.25; cache write $12.50
    Claude Fable 5$10$50per 1M tokens, Cached read $1.00; cache write $12.50
    Claude Opus 5.5$4$20per 1M tokens, Cached read $0.20; cache write $5.00
    Claude Opus 5$5$25per 1M tokens, Cached read $0.50; cache write $6.25
    Claude Opus 4.8$5$25per 1M tokens, Cached read $0.50; cache write $6.25
    Claude Opus 4.7$5$25per 1M tokens, Cached read $0.50; cache write $6.25
    Claude Opus 4.6$5$25per 1M tokens, Cached read $0.50; cache write $6.25
    Claude Opus 4.5$5$25per 1M tokens, Cached read $0.50; cache write $6.25
    Claude Sonnet 5.5$2$10per 1M tokens, Cached read $0.20; cache write $2.50
    Claude Sonnet 5$2$10per 1M tokens, Cached read $0.20; cache write $2.50
    Claude Sonnet 4.6$3$15per 1M tokens, Cached read $0.30; cache write $3.75
    Claude Sonnet 4.5$3$15per 1M tokens, Up to 200K tokens; cached read $0.30; cache write $3.75
    Claude Sonnet 4.5$6$22.50per 1M tokens, Over 200K tokens; cached read $0.60; cache write $7.50
    Claude Haiku 5.5$0.10$0.50per 1M tokens, Up to 100K tokens; cached read $0.01; cache write $0.125
    Claude Haiku 5.5$0.50$2.50per 1M tokens, Over 100K tokens; cached read $0.05; cache write $0.625
    Claude Haiku 4.5$1$5per 1M tokens, Cached read $0.10; cache write $1.25
    Gemini 4 Argon$2$10per 1M tokens, 50% off through November 30, 2026 (was $4.00 input, $20.00 output); cached read $0.10 (was $0.20)
    Gemini 3.8 Flash$1.50$7.50per 1M tokens, Cached read $0.15
    Gemini 3.7 Flash$1.50$7.50per 1M tokens, Cached read $0.15
    Gemini 3.6 Flash$1.50$7.50per 1M tokens, Cached read $0.15
    Gemini 3.5 Flash$1.50$9per 1M tokens, Cached read $0.15
    Gemini 3.5 Flash Lite$0.30$2.50per 1M tokens, Cached read $0.03
    Gemini 3.1 Pro$2$12per 1M tokens, Up to 200K tokens; cached read $0.20
    Gemini 3.1 Pro$4$18per 1M tokens, Over 200K tokens; cached read $0.40
    Gemini 3 Flash$0.50$3per 1M tokens, Cached read $0.05
    Grok 4.7$2$6per 1M tokens, Up to 200K tokens; cached read $0.50
    Grok 4.7$4$12per 1M tokens, Over 200K tokens; cached read $1.00
    Grok 4.6$2$6per 1M tokens, Up to 200K tokens; cached read $0.50
    Grok 4.6$4$12per 1M tokens, Over 200K tokens; cached read $1.00
    Grok 4.5$2$6per 1M tokens, Up to 200K tokens; cached read $0.30
    Grok 4.5$4$12per 1M tokens, Over 200K tokens; cached read $0.60
    Grok Build 0.1$1$2per 1M tokens, Cached read $0.20
    Muse Wonder 1.0$4$20per 1M tokens, Cached read $0.20
    Muse Spark 1.3$1.25$4.25per 1M tokens, Cached read $0.15
    Muse Spark 1.2$1.25$4.25per 1M tokens, Cached read $0.15
    GPT 6 Astra$10$50per 1M tokens, Up to 272K tokens; cached read $1.00; cache write $12.50
    GPT 6 Astra$20$75per 1M tokens, Over 272K tokens; cached read $2.00; cache write $25.00
    GPT 6 Sol$2$10per 1M tokens, Up to 272K tokens; cached read $0.20; cache write $2.50
    GPT 6 Sol$4$15per 1M tokens, Over 272K tokens; cached read $0.40; cache write $5.00
    GPT 6.1 Sol$2$10per 1M tokens, Up to 272K tokens; cached read $0.20; cache write $2.50
    GPT 6.1 Sol$4$15per 1M tokens, Over 272K tokens; cached read $0.40; cache write $5.00
    GPT 6 Luna$0.10$0.50per 1M tokens, Up to 272K tokens; cached read $0.01; cache write $0.125
    GPT 6 Luna$0.20$0.75per 1M tokens, Over 272K tokens; cached read $0.02; cache write $0.25
    GPT 5.6 Sol$4$20per 1M tokens, Up to 272K tokens; cached read $0.40; cache write $5.00
    GPT 5.6 Sol$8$30per 1M tokens, Over 272K tokens; cached read $0.80; cache write $10.00
    GPT 5.6 Terra$2$12per 1M tokens, Up to 272K tokens; cached read $0.20; cache write $2.50
    GPT 5.6 Terra$4$18per 1M tokens, Over 272K tokens; cached read $0.40; cache write $5.00
    GPT 5.6 Luna$0.20$1.20per 1M tokens, Up to 272K tokens; cached read $0.02; cache write $0.25
    GPT 5.6 Luna$0.40$1.80per 1M tokens, Over 272K tokens; cached read $0.04; cache write $0.50
    GPT 5.5$5$30per 1M tokens, Up to 272K tokens; cached read $0.50
    GPT 5.5$10$45per 1M tokens, Over 272K tokens; cached read $1.00
    GPT 5.5 Pro$30$180per 1M tokens, Cached read $30.00
    GPT 5.4$2.50$15per 1M tokens, Up to 272K tokens; cached read $0.25
    GPT 5.4$5$22.50per 1M tokens, Over 272K tokens; cached read $0.50
    GPT 5.4 Pro$30$180per 1M tokens, Cached read $30.00
    GPT 5.4 Mini$0.75$4.50per 1M tokens, Cached read $0.075
    GPT 5.4 Nano$0.20$1.25per 1M tokens, Cached read $0.02
    GPT 5.3 Codex Spark$1.75$14per 1M tokens, Cached read $0.175
    GPT 5.3 Codex$1.75$14per 1M tokens, Cached read $0.175
    GPT 5.2$1.75$14per 1M tokens, Cached read $0.175
    GPT 5.2 Codex$1.75$14per 1M tokens, Cached read $0.175
    GPT 5.1$1.07$8.50per 1M tokens, Cached read $0.107
    GPT 5.1 Codex$1.07$8.50per 1M tokens, Cached read $0.107
    GPT 5.1 Codex Max$1.25$10per 1M tokens, Cached read $0.125
    GPT 5.1 Codex Mini$0.25$2per 1M tokens, Cached read $0.025
    GPT 5$1.07$8.50per 1M tokens, Cached read $0.107
    GPT 5 Codex$1.07$8.50per 1M tokens, Cached read $0.107
    GPT 5 Nano$0.05$0.40per 1M tokens, Cached read $0.005
    Big Pickle$0$0per 1M tokens, Free (page lists Free for input, output and cached read)
    GLM-5.3-Flash Free$0$0per 1M tokens, Free (page)
    DeepSeek V4.1 Flash Free$0$0per 1M tokens, Free (page)
    MiniMax M3.1 Flash Free$0$0per 1M tokens, Free (page)
    Beam Free$0$0per 1M tokens, Free (page)
    LL Alpha Free$0$0per 1M tokens, Free (page)
    LongCat 2.5 Preview Free$0$0per 1M tokens, Free (page)
    Exo Free$0$0per 1M tokens, Free (page)
    MiMo-V2.6-Flash Free$0$0per 1M tokens, Free (page)
    MiMo-V2.5 Free$0$0per 1M tokens, Free (page)
    Nemotron 3 Ultra Free$0$0per 1M tokens, Free (page)
    Nemotron 3.5 Lightning Free$0$0per 1M tokens, Free (page)
    Nemotron 3.5 Super VL Free$0$0per 1M tokens, Free (page)
    Models
    and 78 more
    Works with
    • OpenCode
    • Any coding agent (Zen docs: 'use Console with any coding agent')
    More details
    • Zero markups: the page says you pay what the model provider charges, plus the processing fee (opencode.ai/zen; opencode.ai/docs/zen: 'markup only to cover processing fees')
    • Card processing fee: 4.6% + $0.31 added to each credit purchase; the page shows '+$1.23 card processing fee' on a $20 top-up
    • Auto top-up: when the balance reaches $5, $20 is added automatically
    • Billing can be disabled at any time and the remaining balance can still be used
    • Refund may be available if the charge was made within the last 14 days and the credits from that purchase are unused
    • Gemini 4 Argon 50% discount applies through November 30, 2026
    • Mistral Large 4 is listed at 50% off
    • Rates are quoted from the Zen docs pricing table, which lists per-model input, cached read and output rates; the table's tiered models are split into one row per tier. Nine rows were re-read on a second pass of the same page and matched. The page does not state a markup percentage beyond 'zero markups'.
    Verified 21 Oct 2026Source ↗Zen overview and FAQ ↗Buy on the official site
  • OpenRouter

    Standard (pay-as-you-go)

    Bundle / gatewayPay per tokenClear quota
    Pay per use
    Limits
    Pay-as-you-go: inference is billed per model at the provider's list price, with no markup on inference pricing. The platform fee applies only when buying credits. Free models are limited to 20 requests per minute and 50 per day, rising to 1,000 per day once the account has bought at least $10 of credits.
    Rules
    • If the balance drops below zero, free models are unavailable until funds are added
    • Free models have low rate limits and are usually not suitable for production use
    Works with
    • OpenAI SDK (OpenRouter documents it as a drop-in replacement)
    Worth knowing
    Bring your own key: Standard and Business: $25,000 of list-price inference per month through BYOK with no OpenRouter fee; above that, 5% of what the same request would cost at list price on OpenRouter, deducted from credits. Free plan: BYOK not available.
    More details
    • Platform fee: 5.5% on Standard and 8% on Business, charged when buying credits and never on individual requests; Enterprise has fee discounts available
    • Platform fees are non-refundable
    • Free plan: 25+ free models and 4 free providers
    • Payment by card, AliPay or USDC; the FAQ names one fee on credit purchases (5.5% Standard, 8% Business) and no separate card or crypto fee
    • Business is also pay-as-you-go with an 8% credit fee and the same BYOK allowance; it is not recorded as a separate entry. Enterprise is custom. Models are listed on OpenRouter's model pages and are not enumerated here.
    Verified 21 Oct 2026Source ↗FAQ ↗Quickstart (OpenAI SDK and integrations) ↗Buy on the official site

API only

7 entries, cheapest first

  • Alibaba Cloud (Model Studio / DashScope)

    Model Studio API (pay-as-you-go)

    API onlyPay per tokenClear quotaOff-peak 22:00-08:00 (night), for qwen3.7-plus on Global deployments only UTC+8: Night: 60% off (limited-time); daytime: 20% off (limited-time), as labelled on the page
    Pay per use
    Limits
    Pay as you go, per million tokens, at the rates below. No subscription quota.
    API rates
    ModelInputOutput
    Qwen3.8 Max (0902)$2$6per 1M tokens, Cached input $0.25; cache write $2.50; API name qwen3.8-max
    Qwen3.7 Max$1.475$4.425per 1M tokens, Cached input $0.295; cache write $1.84375
    Qwen3.7 Plus$0.32$1.28per 1M tokens, Cached input $0.064; cache write $0.40
    Qwen3.8 Flash$0.15$0.47per 1M tokens, Cached input $0.016; cache write $0.20
    Qwen3.6 Flash$0.1875$1.125per 1M tokens, cache write $0.234375
    Qwen3.6 Plus$0.325$1.95per 1M tokens, cache write $0.40625

    Rates from OpenRouter, Alibaba's own endpoint, checked 2026-10-21

    Models
    • qwen3.8-max
    • qwen3.7-max
    • qwen3.7-plus
    • qwen3.8-flash
    • qwen3.6-flash
    Worth knowing
    Off-peak: 22:00-08:00 (night), for qwen3.7-plus on Global deployments only (UTC+8), Night: 60% off (limited-time); daytime: 20% off (limited-time), as labelled on the page.
    More details
    • Batch API: 50% discount listed for qwen3.8-max, qwen3.7-max and qwen3.8-flash (Beijing) and qwen3.6-flash (Singapore and Beijing); batch and context-cache discounts cannot be combined
    • Explicit cache creation is billed at 125% of the standard input price (page)
    • Rates are Alibaba's default endpoint on OpenRouter; OpenRouter also lists a us-east-1 endpoint, priced higher for Qwen3.7 Max. The API name qwen3.8-max is served as Qwen3.8 Max (0902).
    Verified 21 Oct 2026Source ↗Model pricing (Chinese page, CNY) ↗Get an API key
  • DeepSeek

    DeepSeek API (pay-as-you-go)

    API onlyPay per tokenClear quotaOff-peak All hours except peak: peak is 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday, excluding Chinese public holidays. Weekends and Chinese public holidays are off-peak all day UTC (Chinese page: peak is 09:00-12:00 and 14:00-18:00 Beijing time, weekdays): 50% (off-peak rates are half of peak rates)
    No subscription
    API only
    Limits
    Pay as you go, per million tokens, at the rates below. No subscription quota.
    API rates
    ModelInputOutput
    DeepSeek V4.1 Flash 1017$0.15$0.60per 1M tokens, Cached input $0.003; API name deepseek-flash; read in DeepSeek's off-peak hours
    DeepSeek V4 Pro 0813$0.66$1.98per 1M tokens, Cached input $0.022; API name deepseek-v4-pro; read in DeepSeek's off-peak hours

    Rates from OpenRouter, DeepSeek's own endpoint, checked 2026-10-21

    Models
    • deepseek-flash
    • deepseek-v4-pro
    Worth knowing
    Off-peak: All hours except peak: peak is 01:00-04:00 and 06:00-10:00 UTC, Monday to Friday, excluding Chinese public holidays. Weekends and Chinese public holidays are off-peak all day (UTC (Chinese page: peak is 09:00-12:00 and 14:00-18:00 Beijing time, weekdays)), 50% (off-peak rates are half of peak rates).
    More details
    • Search and Fetch tools: $0.75 per 1,000 successful requests each; calls made through DeepSeek Harness, Claude Code or Codex are currently free, and per-request billing starts at 00:00 Beijing time on November 1, 2026 (pricing page)
    • Old model names deepseek-v4-flash and deepseek-v4-flash-vision-exp still work and are billed at Flash prices (Chinese pricing page)
    • No subscription or coding plan was found: the pricing page lists per-token billing only. API names: deepseek-flash is DeepSeek-V4.1-Flash-1017 and deepseek-v4-pro is DeepSeek-V4-Pro-0813 (pricing page). OpenRouter's prices for DeepSeek's endpoint were read at 04:46 UTC, inside DeepSeek's off-peak hours, and match DeepSeek's off-peak rates; DeepSeek's page charges twice that at peak.
    Verified 21 Oct 2026Source ↗Pricing (Chinese, CNY) ↗Get an API key
  • Inception Labs

    Inception API (Mercury models, pay-as-you-go)

    API onlyPay per tokenClear quota
    Pay per use
    Limits
    Pay as you go, per million tokens, at the rates below. No subscription quota.
    API rates
    ModelInputOutput
    Mercury 3$0.10$0.35per 1M tokens, Cached input $0.01
    Mercury 2.5$0.04$0.15per 1M tokens, Cached input $0.004
    Mercury 2$0.25$0.75per 1M tokens, Cached input $0.025
    Mercury Decide$0.02$0per 1M tokens, Cached input $0.00

    Rates from OpenRouter, Inception's own endpoint, checked 2026-10-21

    Models
    • mercury-3
    • mercury-2.5
    • mercury-2
    • mercury-edit-2
    • mercury-decide
    More details
    • The models page lists Mercury Router with no price, API name or context length
    • The docs list per-token pricing only, and inceptionlabs.ai/pricing returned 404, so this entry makes no claim about subscriptions. Mercury Voice and Mercury Edit 2 are not listed on OpenRouter, so they have no rate here.
    Verified 21 Oct 2026Source ↗Get an API key
  • Meta

    Meta Model API (pay-as-you-go)

    API onlyPay per tokenClear quota
    Pay per use
    Limits
    Metered API billing: per-token prices for each model, with no minimums or upfront commitment. Standard-mode and Contributor-mode variants have different prices. Rate limits apply per team, and each model has its own limits.
    Rules
    • Contributor variants: Meta may use prompts, code, files, context and model responses to train future models, in exchange for discounted pricing or higher usage
    • Rate limits apply per team, not per API key; for example muse-wonder-1.0 is 3,000 requests per minute and 4,000,000 tokens per minute
    API rates
    ModelInputOutput
    Muse Wonder$4$20per 1M tokens, Cached input $0.20
    Muse Spark 1.3$1.25$4.25per 1M tokens, Cached input $0.15
    Muse Spark 1.2$1.25$4.25per 1M tokens, Cached input $0.15
    Muse Spark 1.1$1.25$4.25per 1M tokens, Cached input $0.15
    Muse Spark 1.3 Contributor$0.10$0.20per 1M tokens, Cached input $0.002; Contributor mode: content is used to improve Meta's products
    Muse Spark 1.2 Contributor$0.10$0.20per 1M tokens, Cached input $0.002; Contributor mode: content is used to improve Meta's products

    Rates from OpenRouter, Meta's own endpoint, checked 2026-10-21

    Models
    • Muse Wonder 1.0
    • Muse Spark 1.3
    • Muse Spark 1.2
    • Muse Spark 1.1
    • Muse Spark 1.3 Contributor
    • Muse Spark 1.2 Contributor
    • Muse Image 1.0
    • Muse Voice Transcribe 1.0
    • SAM 3.1
    Works with
    • Meta Model API (direct API calls)
    More details
    • Standard mode: content is not used to train Meta models
    • Web search grounding: $2.50 per 1,000 search queries, charged in addition to token costs
    • Muse Image (muse-image-1.0): $0.01 per generated image
    • Muse Voice Transcribe (muse-voice-transcribe-1.0): $0.18 per hour of audio, metered in whole seconds
    • SAM 3.1 (sam-3.1): $2.50 per 1,000 images or $0.20 per 1,000 video frames
    • Muse Wonder Full 1.0 has no API price on the page; it is sold only by subscription in Muse Code (see the Muse Code entries). The product page lists no regions and does not say whether the API is in public preview or generally available.
    Verified 21 Oct 2026Source ↗Meta Model API product page ↗Get an API key
  • MiniMax

    MiniMax API (pay-as-you-go)

    API onlyPay per tokenClear quota
    Pay per use
    Limits
    Pay as you go, per million tokens, at the rates below. No subscription quota.
    API rates
    ModelInputOutput
    MiniMax M3$0.30$1.20per 1M tokens, Cached input $0.06

    Rates from OpenRouter, MiniMax's own endpoint, checked 2026-10-21

    Models
    More details
    • Video (pay-as-you-go, billed per second): MiniMax-H3 768P $0.08, 2K $0.13; MiniMax-H3-Max 480P $0.05, 768P $0.08; H3 regeneration 768P to 2K $0.05 per regenerated second
    • Video input images (H3): first 5 free, then $0.04 each; H3-Max: first 2 free, then $0.074 each
    • Speech (T2A, per 1M characters): speech-2.8-hd $100; speech-2.8-turbo $60; asynchronous long-form synthesis up to 1M characters per request
    • Voice design $3 per voice and rapid voice cloning $1.50 per voice, charged on first synthesis use
    • Speech-to-text (ASR) $0.38 per hour
    • Legacy per-video Hailuo models are also listed (for example MiniMax-Hailuo-2.3 768P 6s $0.28)
    • Video is not API-only: the M Plan Explore and Build subscriptions include H3 video. Video is priced per second in the API; text is priced per 1M tokens. MiniMax M3.1 Flash is listed on OpenRouter only as a free variant, so it has no paid rate here.
    Verified 21 Oct 2026Source ↗Get an API key
  • Moonshot AI (Kimi)

    Kimi API (pay-as-you-go)

    API onlyPay per tokenClear quota
    Pay per use
    Limits
    Pay as you go, per million tokens, at the rates below. No subscription quota.
    API rates
    ModelInputOutput
    Kimi K3$3$15per 1M tokens, Cached input $0.30
    Kimi K2.7 Code$0.95$4per 1M tokens, Cached input $0.19
    Kimi K2.7 Code (highspeed)$1.90$8per 1M tokens, Cached input $0.38; highspeed endpoint
    Kimi K2.6$0.95$4per 1M tokens, Cached input $0.16

    Rates from OpenRouter, Moonshot AI's own endpoint, checked 2026-10-21

    Models
    More details
    • Billing is by usage; the platform states it does not offer subscription plans and is separate from Kimi membership and Kimi Code
    • Web search is charged an additional 0.03 CNY per call that uses it (Kimi API pricing page)
    • The Kimi API platform says it sells no subscription plan; Kimi Code memberships are separate entries.
    Verified 21 Oct 2026Source ↗API pricing guide ↗K3 launch post ↗Get an API key
  • Xiaomi (MiMo)

    MiMo API (pay-as-you-go)

    API onlyPay per tokenClear quota
    Pay per use
    Limits
    Pay as you go, per million tokens, at the rates below. No subscription quota.
    API rates
    ModelInputOutput
    MiMo-V2.6-Pro$0.435$0.87per 1M tokens, Cached input $0.0036
    MiMo-V2.6-Flash$0.14$0.28per 1M tokens, Cached input $0.0028
    MiMo-V2.6-Pro-UltraSpeed$4.35$8.70per 1M tokens, Cached input $0.036

    Rates from OpenRouter, Xiaomi's own endpoint, checked 2026-10-21

    Models
    More details
    • No time-based token discount is listed on the pay-as-you-go page; cache writes are described as limited-time free with no end date
    • TTS models (mimo-v2.5-tts and variants) are free for a limited time and are not token-priced
    • mimo-v2.5-pro and mimo-v2.5 retired at 10:00 Beijing time on October 21, 2026, per the price page. OpenRouter lists no batch rates, so batch prices are not shown.
    Verified 21 Oct 2026Source ↗Get an API key