🚀 Anthropic launches Claude Fable 5.2 and Mythos 5.2 🚀
Same model, two access levels: faster, cheaper and stronger at coding.

Anthropic has upgraded its top-tier models with Claude Fable 5.2 and Claude Mythos 5.2. They are the same underlying model offered in two versions. Fable ships with strong cybersecurity and biology safeguards and is available to all users. Mythos is reserved for vetted organizations through Anthropic's verification programs. The new models generate output up to 30% faster than version 5.1 and cost less, at $8/$40 per million input/output tokens. On FrontierCode 2.0, Fable 5.2 posts the highest score of any released model tested. At High effort it matches Opus 5.5's best result for roughly half the cost. In prerelease testing it solved two open problems in black-hole physics and formalized Fermat's Last Theorem in 30 hours with about 1,000 agents. For everyday work, Anthropic still recommends Opus 5.5 as the default model.

🕵️ Nvidia's Jensen Huang admits affair as board orders review 🕵️
Spy claims, an independent probe and a Nov. 11 deadline from Congress.

Nvidia CEO Jensen Huang has admitted to an extramarital affair in a letter to staff. This follows reports that he was targeted by a foreign intelligence effort through a customer executive whom US officials reportedly linked to China's spy agency. Nvidia's board has launched an independent investigation, overseen by independent directors with outside counsel. The board says it has no reason to believe national security or company secrets were compromised, and Huang remains CEO. China denies the spying claims. Pressure is also building in Washington: Huang has until Nov. 11 to answer House Republicans' request for a closed-door briefing, proposed for Nov. 18. For the world's key AI chip supplier, the case adds political risk at a sensitive moment for US-China chip policy.

🎨 Meta adds custom Spaces to its Muse app 🎨
Describe what you need, and Muse builds it.

Meta has added Spaces to its Muse app, letting users build personalized interactive experiences just by describing what they want. Examples range from gift trackers and project planners to small games, all created in natural language rather than code. The move turns Muse into a home for lightweight mini-apps built around everyday needs. It also brings vibe-coding-style creation to a mainstream consumer audience usually served by specialized builder tools. Given Meta's reach, Spaces is a strong distribution play and a clear signal that big platforms want to own the space where non-developers turn ideas into working tools.

INTERESTING TO KNOW

🧠 Anthropic shows how AI agents build memory by 'dreaming' 🧠

Anthropic has published research on how Claude agents build long-term memory. They save notes as files while they work, then use idle periods to check those notes against logs of past tasks, fixing errors and reorganizing what they keep. Anthropic calls this process "dreaming". The company found that newer models are much better at judging what is worth remembering, a step toward agents that improve the longer they work.

🔀 OpenRouter launches new Auto Router and Auto Code 🔀

OpenRouter has rebuilt its Auto Router, which picks the best model for each request, and added Auto Code, a version tuned for coding agents. The routers weigh eleven factors per request alongside live pricing, provider health and cache status. They also respect guardrails and zero data retention rules. OpenRouter says Auto Code scores in line with Claude Opus 5.5 across five benchmarks while costing 74% less per task, with no extra routing fee. Auto Code is not yet available through EU in-region routing.