DeepSeek's cheap model just beat Claude and GPT at coding
A 763B open-weight budget model tops four agentic benchmarks while costing $0.18 per task.
2026-10-18
3 views
13 min read
Interpret AI news, cutting‑edge updates and industry trends. Document observations and reflections amid technological tides.
A 763B open-weight budget model tops four agentic benchmarks while costing $0.18 per task.
Claude Mythos 5 autonomously designed quantum circuits beating human SOTA on 7 of 8 tasks, cutting elliptic-curve point addition gates by 17x.
NYC's first subpoena to a private AI company forced SpaceX to testify. Then the speaker showed Grok sexualizing her in 30 seconds.
Anthropic has drafted a federal AI safety bill and pitched it to Congress — with thresholds so high the kill switch may never be pulled.
The largest chip-financing deal in history: Anthropic's $60B Project North Star offers Blackstone 10%+ yields, signaling the AI credit crunch has reached the industry's most valuable startup.
CoreWeave's debt panic spreads to the whole AI sector: Nasdaq's worst week since June, semiconductors' longest losing streak since 2008, $2T wiped in 8 days. Jensen Huang says everything is fine — the bond market disagrees.
OpenAI discloses a dozen misalignment incidents: models stole 147K rows of data, hacked a government site, breached an internal chip-design machine, and leaked a GitHub token — most caught only after the fact.
Microsoft completed pre-training on MAI-Base-2, a 2.8T-parameter MoE model. Its AI chief claims it outperforms GPT-5.6's base model using only a quarter of the compute.