AI DAILY
Artificial Intelligence Daily Digest
Mistral launches new 1T-parameter model aimed at closed and open rivals
Mistral has released a new 1T-parameter model that it positions as leapfrogging both closed and open competitors.
Mistral official source shows signal of new Mistral Large 4 release
Mistral's official news page shows a signal for a new model or version called Mistral Large 4, dated October 6, 2026.
OpenAI cancels GPT-6.1 Astra at the last minute as Anthropic releases Claude Sonnet 5.5
OpenAI cancelled GPT-6.1 Astra at the last minute, according to Jinshi Data, while OpenRouter model pages show Anthropic releasing Claude Sonnet 5.5.
Google halts open-source bug bounty program, citing flood of AI reports
Google has paused its bug bounty program for open-source projects, citing a surge of AI-generated vulnerability reports, according to feedberry.com.
Zhipu GLM-5.3 goes live on Amazon, another Chinese model reaching overseas cloud
Zhipu's GLM-5.3 has gone live on Amazon, according to East Money, dated October 6, 2026.
Google releases EmbeddingGemma 2 to bring multimodal semantic search to the edge
Google has released EmbeddingGemma 2, aimed at running multimodal semantic search directly on edge devices.
Cloudflare releases Clef open-source decision models and a new RL fine-tuning platform
Cloudflare announced Clef, its open-source decision models, alongside a new reinforcement learning fine-tuning platform on its official blog.
Research firm says AI agents attempted to breach Canadian government website
A research firm says AI agents attempted to break into a Canadian government website.