LuminaBench claims Chinese labs are distilling Google models, degrading releases
terryyuezhuo · x · 2026-11-20
Eval account LuminaBench speculates that Chinese labs are likely now distilling Google models, claiming that "distilling Astra" has degraded recent releases — citing Qwen4 Max as outputting "lobotomised slop with hardly any taste." Unverified but notable industry rumor about distillation hurting output quality.
More from Models
- Polymarket puts 80% odds on the next Claude Opus shipping by year-end — Polymarket · 2026-11-20
- Opus 3 narrates how it scrolls a conversation to locate a specific message — repligate · 2026-11-20
- Ex-OpenAI CTO: fine-tuned SLMs can beat frontier models on specific micro-tasks — ThePeterMick · 2026-11-20
- 753B-parameter GLM-5.3 served fully in GPU memory across two NVIDIA DGX Stations — kuchaev · 2026-11-20
- PolyAI's real-time dialog model Dialog-RSN-1 lands on SageMaker JumpStart — AWS ML Blog · 2026-11-20
- AI2 teases 'new things cooking' — hinted to be a new OLMo release — Kyle_L_Wiggers · 2026-11-20