Speculation: top open-weight models may be distilling OpenAI and Anthropic, missing training code hints
dan_s_becker · x · 2026-09-16
dansbecker speculates that if top open-weight models are largely distilling OpenAI and Anthropic, frontier labs could pace the frontier without cooperation from Chinese labs. He notes open-weight models shipped without training code are much better than those with code — consistent with heavy distillation. Unverified speculation.
More from Models
- Dev warns: Claude Code's 1M context window can burn through your budget fast — peterjliu · 2026-09-16
- Essay claims Claude's personality converges on Anthropic's company culture — ryunuck · 2026-09-16
- ChatGPT Co-Inventor's New AI Startup Claims 200x Speed, Free Output Tokens Forever — Polymarket · 2026-09-16
- GPT-6 debugs Fallout 3's infamous Sentinel Lyons dialogue glitch — imjustnewatai · 2026-09-16
- That 1M-token context window can really burn your bill — peterjliu · 2026-09-16
- Science LLM benchmarks have flawed answers; fixing them significantly raises model scores — Profanion · 2026-09-16