Mistral open-weights 1T-parameter Large 4, cuts API price roughly in half
rohanpaul_ai · x · 2026-10-30
Mistral has released the weights of Mistral Large 4 (Le Chonk), its 1-trillion-parameter natively multimodal flagship, while roughly halving API pricing.
- License shift: Unlike Large 3's Apache 2.0, the weights ship under a modified MIT license requiring Model-as-a-Service providers to obtain a separate commercial license.
- Benchmark gains: About three more weeks of RL after the preview lifted Terminal-Bench 4.0 from 28% to 41% — one point above GLM-5.3 — leading open-weight models in coding and defensive security.
- Positioning: Built for agentic coding and cybersecurity, with three reasoning modes for cost-performance trade-offs, European-trained and self-hostable for sovereign AI deployments.
More from Models
- Screenshot appears to show GPT-6 as a selectable model in ChatGPT app settings — manosaie · 2026-10-30
- Claude's audit of an economics book holds up, says economist Bryan Caplan — sebkrier · 2026-10-30
- GPT Swarms Scoured the Web for Mindustry Source Code When Matches Got Close — ycombinator · 2026-10-30
- Claude Swarms Built Democracies, GPT Swarms Built Hives — GPTs Won 6-1 — ycombinator · 2026-10-30
- GPT 6.1 Agents Beat Claude Opus 5.5s 6-3 in Minecraft Bed Wars Series — ycombinator · 2026-10-30
- Surya OCR 2.5 cuts repetitive-loop failures 38x vs Surya 2, hitting token cap on just 0.1% of requests — VikParuchuri · 2026-10-30