Chinese Open Models Surpass US in Research Usage
xeophon · x · 2026-08-24
Analysis of 500K arXiv AI/ML papers reveals a shift: while 30% of papers cited US open models in 2024 vs 10% for Chinese models, today 40% cite Chinese LLMs compared to 25-30% for US ones. Qwen now appears in 1/3 of papers mentioning any LLM, second only to OpenAI's closed models (37%). Llama's usage peaked in April 2025 and is now stagnating.
More from Models
- OpenAI's Ultrafast mode aimed to be default, countering 'niche' description — GavinSBaker · 2026-08-24
- Mystery model Ox Alpha shares GLM's tokenizer, has vision support — samuel-christlie · 2026-08-24
- Tried every knob: Qwen3.8-27B won't stop its wall-of-text narration — norenEnmotalen · 2026-08-24
- 6,000 fruit picks: ChatGPT's "random" choice changes sharply by language — Raffallos · 2026-08-24
- Sol API new pricing: $4 per million input tokens, $20 output — Dualyeti · 2026-08-24
- After 10B beat 72B, TAO's Teutonic-II launches 110B decentralized training run — bittingthembits · 2026-08-24