Mistral Large 4 Preview: 1T-Parameter MoE Tops All Non-Chinese Open Models
The Batch (Andrew Ng) · rss · 2026-10-16
Mistral released an API preview of Mistral Large 4 ("le Chonk"), its largest model at 1.05T parameters (52B active, MoE) with a 1.6B vision encoder, trained from scratch on 3,800 GPUs across 160+ languages; weights open by end of month. It scores 38 on the Intelligence Index, tying GPT-6 Luna, behind three cheaper Chinese open-weight models but ahead of all US open models. Its standout is security: 82% on CyberGym-E2E-AA memory-safety bug fixing, best among public models. Priced at $1.36/$0.14/$4.18 per M tokens with launch and batch discounts. Training used a single continuous RL run mixing chat, science, safety, and long-horizon tool-use tasks with rolling model updates.
More from Models
- OpenAI's Decisions API hits GA: near real-time routing up to 10x faster than GPT-6 Luna — OpenAIDevs · 2026-10-23
- DeepLearning.AI weekly: GPT-6.1 hits 97.9% byte parity on COBOL migration — DeepLearningAI · 2026-10-23
- Marker's lite mode scores 75.1 vs 42.3 for Azure Read on OmniParseBench — VikParuchuri · 2026-10-23
- Datalab launches lite OCR at $0.50 per 1,000 pages, 20-30x cheaper than Azure, Google and AWS layout APIs — VikParuchuri · 2026-10-23
- LiveNerf's 29-day data finds no nerfing of Claude Opus 5.5 — TheOnlyVibemaster · 2026-10-23
- PHP Foundation reveals Anthropic's Claude cyber model scanned PHP, validated with OpenAI's Daybreak Red — saibharadwaj · 2026-10-23