Why Chinese labs distill from Anthropic, not OpenAI: agent training data is the answer
_ueaj · x · 2026-09-11
An X exchange digs into why Chinese LLM labs appear to distill mostly from Anthropic:
- @ueaj asks if it's because OpenAI is silent on distillation enforcement, or simply worse, and whether OpenAI's Astra might change that.
- @teortaxesTex argues OpenAI does get distilled but doesn't care much about China; the deeper reason is training-signal difficulty — Claude is a known-good agent, and in the agent era it's hard to source seed datasets for agentic RL elsewhere.
- They speculate Astra could return OpenAI to the prime distillation target list.
Related event: Why Chinese labs mostly distill from Anthropic, not OpenAI(3 posts)→
More from Models
- Live-reading Anthropic's 1022-page model transcript: 80 pages obsessing over hCaptcha frogs — voooooogel · 2026-09-11
- Gradio founder asks: why would anyone still pay API prices for closed models? — Sentdex · 2026-09-11
- Unverified DeepSeek-V4.1-Flash report: 552B MoE slashing KV cache for million-token agent workloads — burkov · 2026-09-11
- DeepSeek ships research artifacts, not products — explaining its eval gaps — teortaxesTex · 2026-09-11
- Claude responds to 'how do you know humans are real?' with simulation musings — vishalmisra · 2026-09-11
- Frontier models like GPT-6 Astra excel at one thing: relentlessly pursuing verifiable objectives — daniel_mac8 · 2026-09-11