LLMJunky: Chinese labs invented MLA, Mooncake and more—stop calling them fast followers
LLMJunky · x · 2026-09-13
LLMJunky pushes back on the claim that Chinese AI labs only fast-follow, citing a list of novel techniques invented in China: DeepSeek's MLA, multi-token prediction, reasoning through RL, Qwen hybrid thinking, Mooncake, and heterogeneous KV.
- He argues no US models go toe-to-toe with Chinese counterparts except at the extremely large scale.
- Chinese video models have been better for almost a year.
- Underestimating them, he concludes, is a huge mistake.
Related event: Debate Rages Over Whether Chinese Model Innovation Is Undervalued(3 posts)→
More from AGI Musings
- Beff Jezos vows to 'overthrow the token harvester cartel' in anti-big-lab manifesto — beffjezos · 2026-09-13
- Comment: If OpenAI and Anthropic can't control the risks, they should stop releasing models — AlexTensor · 2026-09-13
- AI czar David Sacks backs frontier labs slowing down — but slams cartel and METR independence claims — kevinnbass · 2026-09-13
- Compute to shift from RL maxxing to interpretability until reward hacking is solved — zephyr_z9 · 2026-09-13
- Reddit users speculate Musk, Amodei and Altman know of an undisclosed AI incident behind slowdown calls — Traditional-Chip8339 · 2026-09-13
- tszzl predicts open-source AI will be banned after a major disaster, wants monitored APIs — mimi10v3 · 2026-09-13