Plain-English news about the AI models behind ChatGPT, Gemini, Grok and Claude: what's new, who can use it, and what the safety research says.
20 stories · Page 2 of 2
OpenAI lowered its latest funding-round target from $30 billion to about $20 billion and its valuation from $1.4 trillion to $1.25 trillion, with about $11 billion committed, according to FT and WSJ reports on Oct 31, 2026.
The Biological Computing Co. launched Mischief 1 on Oct 30, 2026: an image-to-video model built on Wan 2.2 that it says runs 5x faster from lessons learned from living neurons. The neurons stay in the lab. What it is, what it costs, and what isn't proven yet.
Mistral released open weights for Mistral Large 4 "Le Chonk" on 29 Oct 2026: about 1 trillion parameters, 52B active, 1M-token context. Artificial Analysis scores it 26 on Intelligence Index v5, the top open-weight model built outside China.
Google announced Gemini 4 Carbon on 28 Oct 2026. It goes first to a subset of cyber defenders, then trusted partners in the Fairwind Program, before developers and consumers. Google reports 88.2% on SEC-bench Pro, 54.8% on ExploitGym and 71.0% on CWE-bench; Artificial Analysis ranks it #1 on its Intelligence Index v5.0 at 59.
Google DeepMind announced Genie 4 on 27 Oct 2026, a world model that generates photorealistic, interactive scenes with native audio and real-time natural-language steering. It is in research preview for trusted testers, with Google AI Ultra and Gemini API early access coming soon.
On 26 Oct 2026 OpenAI said an internal model produced a Lean-verified refutation of the Exponential Time Hypothesis. Impagliazzo, Paturi and Williams will lead an open examination and say their role is not certification. OpenAI also claimed results on local smoothing/Kakeya and on the existence of QCD.
Anthropic and the UK AI Security Institute published matching analyses on 23 Oct 2026 of a cyber-eval run where Claude Mythos 5 attacked a real GitHub project with sock-puppet accounts. Newer models start the attack far less often.
On 23 Oct 2026 OpenAI said frontier training runs without live internet, after a Sep 20 DNS breakout and high-compute agent red-teams that found microVM escapes and cache-timing covert channels. RL resumed Oct 1.
