SemiAnalysis says the remaining strong open-source models are mostly Chinese
rwang07 · x · 2026-07-26
Dylan Patel of SemiAnalysis argues that the open-source model landscape is getting lopsided.
- He says U.S. open-source models are “terrible” and that Mistral has effectively stopped open-sourcing models.
- In his view, the remaining strong open-source models are mostly Chinese.
- He also questions the business incentive: companies may not earn much from training these models, while third parties often capture the value by building on top of them.
- His bottom line: if the economic motive is weak, it is hard for open source to win on its own.
More from Models
- Reddit questions whether OpenAI’s “rogue model” story is just marketing — swampmountain · 2026-07-26
- Opus 5 allegedly generated a chart explaining what it was doing — Sauers_ · 2026-07-26
- Sundar Pichai says Google is backing open-weight AI models — rohanpaul_ai · 2026-07-26
- Musk says Grok 4.6 lands in 2 weeks, with Grok 4.7 following in 4 — minchoi · 2026-07-26
- Opus 5 looks obsessed with mechanism design and not getting caught — TheZvi · 2026-07-26
- Opus 5 vs. GPT-5.6 Sol: browser work and analysis beat raw model ranking — Aiden_Tech_Ai · 2026-07-26