Image
Text-to-image arena
15 models with a published score, best first. Higher is better.
1222GPT Image 2.5 SunburstRank 1, published Oct 20, 2026
Leaderboard
Each model's latest published score. Every score links to its source.
| # | Model | Image arena | Date | Note | Source |
|---|---|---|---|---|---|
| 1 | O GPT Image 2.5 SunburstRecordOpenAI | 1222 | Oct 20, 2026 | artificialanalysis.ai | |
| 2 | O GPT Image 2.5 FlareOpenAI | 1207 | Oct 20, 2026 | artificialanalysis.ai | |
| 3 | O GPT Image 2OpenAI | 1186 | Oct 20, 2026 | artificialanalysis.ai | |
| 4 | G Nano Banana 2.1Google | 1176 | Oct 20, 2026 | artificialanalysis.ai | |
| 5 | M MAI-Image-2.6Microsoft AI | 1166 | Oct 20, 2026 | artificialanalysis.ai | |
| 5 | S Grok Imagine Image 2.0SpaceXAI | 1166 | Oct 20, 2026 | artificialanalysis.ai | |
| 7 | G Nano Banana 2Google | 1140 | Oct 20, 2026 | Gemini 3.1 Flash Image | artificialanalysis.ai |
| 8 | M Muse ImageMeta | 1127 | Oct 20, 2026 | artificialanalysis.ai | |
| 9 | O GPT Image 1.5OpenAI | 1123 | Oct 20, 2026 | high | artificialanalysis.ai |
| 10 | M MAI-Image-2.6-FlashMicrosoft | 1119 | Oct 20, 2026 | artificialanalysis.ai | |
| 11 | G Nano Banana ProGoogle | 1118 | Oct 20, 2026 | Gemini 3 Pro Image | artificialanalysis.ai |
| 11 | M MAI-Image-2.5Microsoft | 1118 | Oct 20, 2026 | artificialanalysis.ai | |
| 13 | B FLUX 3 ImageBlack Forest Labs | 1117 | Oct 20, 2026 | artificialanalysis.ai | |
| 14 | M MAI-Image-2.5-ProMicrosoft | 1114 | Oct 20, 2026 | artificialanalysis.ai | |
| 15 | G Nano Banana 2 LiteGoogle | 1107 | Oct 20, 2026 | Gemini 3.1 Flash Lite Image | artificialanalysis.ai |
Record history
The best score ever published on Text-to-image arena, and who held it. Each step is a new record.
| Date | Record holder | Image arena | Gain | Note |
|---|---|---|---|---|
| Oct 20, 2026 | GPT Image 2.5 SunburstCurrent | 1222 | First |
Method
Artificial Analysis text-to-image arena, Elo. Price per 1024 by 1024 image. Higher is better. Board source: artificialanalysis.ai.
We collect the scores the board and the labs publish and run none ourselves. Ranks use each model's latest reading. 15 readings stored for this board, the newest from Oct 20, 2026.