AI Daily · Oct 9, 2026: GPT-6.1 Sol Ultrafast launches with up to 8x speed
OpenAI launches GPT-6.1 Sol Ultrafast with up to 8x faster speeds and instant steering, live on API, Codex, and ChatGPT Work. Meanwhile Google releases on-device multimodal embedding model EmbeddingGemma 2, Qwen open-sources 8-step Image 2.1 Turbo, and Step 5 Preview plus Qwen models offer limited-t
Pick a date
-
OpenAI Releases GPT-6.1 Sol Ultrafast With Faster Steering
OpenAI's Tibo announced GPT-6.1 Sol Ultrafast is rolling out to the API, Codex, and ChatGPT Work, delivering up to 8x faster speeds than Sol Standard alongside instant steering adjustments.
-
Google releases EmbeddingGemma 2 for on-device multimodal embeddings
Google DeepMind introduced EmbeddingGemma 2, an open embedding model for on-device use supporting code, image, video, audio, and text embeddings, with modular footprints ranging from 270M for text/code up to 740M for full multimodal.
-
Qwen releases open-weight Image 2.1 Turbo with 8-step generation
Alibaba Qwen released Qwen-Image-2.1-Turbo with open weights: an accelerated checkpoint of the 7B Qwen-Image-2.1 architecture that generates and edits images in 8 denoising steps while still producing 2K-resolution output, runnable locally with Diffusers.
-
Tencent open sources Hy-MT2 offline translator supporting 33 languages
Tencent's Hy-MT2 is an open-source translator weighing 440 MB that runs offline on mobile devices and supports 33 languages.
-
Step 5 Preview reaches OpenRouter leaderboard and Kilo free tier
StepFun's Step 5 Preview appeared on OpenRouter's New & Trending leaderboard, and Kilo is offering it free for a week; the model is described as 600B parameters with 27B active per token, 1M-token context, and vision support.
-
Qwen offers week of free access on GMI Cloud with raised limits
Qwen is providing a week of free access to Qwen3.8-Max, Qwen3.8-Flash, and Wan3.0 on GMI Cloud, and the free tier for Qwen 3.8 Max and Flash has been extended by an additional 7 days with increased rate limits across all three models.
-
Grok Imagine Video 1.5 Lite Launches In API With Pricing
The Grok Imagine API now offers Video 1.5 Lite for text-to-video and image-to-video generation at $0.02/sec for 480p, $0.03/sec for 720p, and $0.14/sec for 1080p.
-
Anthropic pauses Claude Startups Team plan and $1,000 API credits
Due to hundreds of thousands of applicants, Anthropic is pausing the Claude Team plan and $1,000 API credit benefits in the Claude Startups program while it re-reviews applications; in the first 48 hours it received 21 times the applications seen over the previous 5 months.
-
Anthropic launches Cyber Mission and free OSS Scanner vulnerability service
Anthropic announced the Anthropic Cyber Mission to protect critical infrastructure and open-source software, including a free OSS Scanner that uses its frontier models to periodically scan opted-in projects for vulnerabilities, delivering proof-of-concept, explanation, and fix suggestions.
-
Anthropic usage policy may ban users for abusive behavior toward Claude
Anthropic's upcoming Usage Policy would prohibit sustained and needless abusive or cruel behavior toward its models, potentially allowing the company to ban users who bully Claude.