Mistral open-weights Le Chonk: 1T-param MoE with 52B active, 1M context, best non-China open model
mark_k · x · 2026-10-30
Mistral has released open weights for Le Chonk (Mistral Large 4): a 1T-parameter MoE with 52B active parameters, a 1M-token context window, image input and three reasoning modes. It scores 26 on Artificial Analysis, making it the strongest open model built outside China — though still 6 points behind DeepSeek V4.1 Flash and GLM-5.3. Cost per task dropped 63% versus the preview, with API pricing at $0.70/M input and $2.10/M output tokens.
More from Models
- GPT-6.5 'Bel' may be pricey but could make AI cheaper by training better small models — VraserX · 2026-10-30
- Reddit meme: labs will rush releases before Qwen 4 just to top benchmarks for a day — Porespellar · 2026-10-30
- HF sits down with Mistral team for a deep dive into building Large 4 — SergioPaniego · 2026-10-30
- Humansand launches &Tutor, a tutor model trained by teaching a human-like learner model — techNmak · 2026-10-30
- Benchmark sweep shows greedy decoding tanks coding accuracy: 70% at temp 0 vs 82.7% at temp 1.0 — Brief-Tap-6616 · 2026-10-30
- Gemini 4 tops public coding benchmarks but drops to bottom on private tasks — Hesamation · 2026-10-30