Mistral releases Mistral Large 4: open-weight 1T-param MoE with 52B active, trained on 3800 Blackwell GPUs
jacek2023 · reddit · 2026-10-29
Mistral released Mistral Large 4 ("Le Chonk") on Hugging Face, its largest and most capable model:
- 1 trillion parameters, natively multimodal MoE with 49-52B active params per token
- Trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own European datacenters
- Unifies instruction following, reasoning, coding and agentic workflows in one set of weights; strong on cybersecurity, finance and law, beats frontier closed models on visual grounding
- Natively fluent in 160+ languages, including every official EU language
Also served via the Mistral API as mistral-large-4.
Related event: Mistral Open-Sources Large 4, Its 1T-Parameter Flagship Model(3 posts)→
More from Infra
- Oracle plans 10GW capacity by FY28 as AI backlog jumps 4.6x to $638B in one year — rohanpaul_ai · 2026-10-29
- Unsloth quants run 1T-param Mistral Large 4 on Mac Studio: 1-bit keeps 91% accuracy at 223GB — danielhanchen · 2026-10-29
- SMCI-backed neocloud Argentum AI seeks $5B credit facility for chip purchases — Beth_Kindig · 2026-10-29
- Microsoft returns to bond market seeking $25B+; AI debt wave lifts borrowing costs for all — rohanpaul_ai · 2026-10-29
- Self-hosted realtime avatar server: 30 FPS to 90 FPS on a free Colab T4 — SavingsWeather1659 · 2026-10-29
- WSJ: Apple refused flash deliveries when prices fell, leaving suppliers stuck — firstadopter · 2026-10-29