ThursdAI weekly: OpenAI claims 10K-agent swarm solved Navier-Stokes, DeepSeek V4.1 shrinks KV cache to ~890 bytes/token
altryne · x · 2026-09-12
- OpenAI claims Navier-Stokes breakthrough: a swarm of 10K agents running an unreleased OpenAI model reportedly solved the millennium math problem; the show debates credibility.
- DeepSeek V4.1's KV cache compression: Nvidia's Chris Alexiuk breaks down DeepSeek's compression path down to 890 bytes per token — cache hits vs misses dominate the agent bill, letting long context stay hot without GPU cost exploding; frontier-level capability at a fraction of the price.
- Meta launches Muse: a free AI agent with its own computer, positioned to replace OpenClaw/Hermes-style tools for most users.
- The Anthropic quitter's doomer essay: "AI is going to kill all of us" hit 130M views on X, sparked TV interviews, revived the doomerism movement, and reportedly led OpenAI to consider pausing training.
- Also: a week with Google Astra ("not quite AGI"), and a surprise CoreWeave FullyConnected announcement. 1h43m podcast with full transcript.
Related event: ThursdAI Recap: 10,000-Agent Millennium Claim and AI Safety Week(8 posts)→
More from Companies & People
- Silicon Valley's acqui-hire gold rush: 9-month startups, instant millions, and shrinking ambition — edgarpavlovsky · 2026-10-27
- Sam Altman to join John Collison for SF fireside chat on SaaS and AI — jeff_weinstein · 2026-10-27
- Robot learning researcher joins Mila to work on physical AI and RL — pcastr · 2026-10-27
- The Information Distills Key Takeaways from Anthropic's Financials — steph_palazzolo · 2026-10-27
- AI cycle is entering its 'late-stage crypto' phase, argues investor: minor labs fade, startups pivot to GPU inference clouds — beffjezos · 2026-10-27
- Ex-Anthropic employee essay: the org optimizes for power 'constrained by being seen as the good guys' — RichardMCNgo · 2026-10-27