Matching stories Newest first
Elevenlabs releases two Eleven Realtime speech-to-speech models
The new Eleven Realtime speech-to-speech models target real-time AI conversation. The v1 Max variant scores 98.8 percent on Big Bench Audio with a 0.57-second response time, according to Elevenlabs, which claims no model on the Artificial Analysis leaderboard is both faster and more accurate.
Hugging Face model Spark-X2.6-4B, a compact on-device language model, is released
Spark-X2.6-4B keeps the Spark-X2.5-4B architecture and gains from improved post-training, with three thinking levels (none, low, high). The developer claims it matches or beats models around 9B parameters on open-source benchmarks.
Gradium releases Phonon, open-source on-device text-to-speech models
Gradium released Phonon, two small TTS models of 31M and 40M parameters, with a fully open-source inference stack targeting CPUs, GPUs and NPUs across 10+ backends, mostly written in Rust.
Ornith-1.5-122B, a Qwen-based finetune, released in GGUF and NVFP4 formats
Ornith AI introduces Ornith-1.5, which continuously generates its own training tasks and improves through reinforcement learning, extending Ornith-1.0 built on Qwen3.5 and Gemma4. GGUF and NVFP4 builds are on Hugging Face, with more details in the company blog.
Cloudflare releases Clairic, a model that extracts data from documents
Cloudflare says Clairic pulls data from documents faster and more cheaply than large language models.
OpenAI prepares GPT-6.1 Terra, with a cheaper Terra model reportedly in the works
GPT-6.1 Terra appeared in a Microsoft pull request, and The Information reports OpenAI is readying a cheaper Terra model. Neither report includes an official launch date.
Kling AI releases Kling 4.0 video generator with stereo audio and lip sync
Kling 4.0 supports clips up to 30 seconds, ten keyframe images as input, and prompts of up to 8,000 tokens, plus a lighter Flash variant. Comparisons with Wan 3.0 and Seedance 2.5 show it holding its own without clearly pulling ahead.