Mistral Large 4
Mistral's trillion-parameter mixture-of-experts flagship, in API preview now with open weights promised this month.
Quick verdict
- Best for
- Teams needing a European or self-hostable flagship
- Pricing
- $1.36 / $4.18 per million tokens (launch sale $0.68 / $2.09); cached $0.14
- Not ideal if
- Weights not yet downloadable at preview
What is Mistral Large 4?
Mistral Large 4 entered public preview on October 6, 2026 as a granular mixture-of-experts model of about 1 trillion total parameters with roughly 49 to 52 billion active and a 1.6B vision encoder, priced at $1.36 per million input tokens and $4.18 per million output with a launch sale at half those rates. Mistral says open weights will follow by the end of October; at launch it is API-only. Reported context is 1M tokens by Mistral's spec sheet, with independent evaluators measuring about 524K and a 262K output limit.
Key features
- ~1T total / ~50B active mixture of experts with vision
- Tool calling for agent loops
- 1M-token context per Mistral (independently measured ~524K)
- Open weights promised by end of October 2026
- European provider with EU hosting
Pros
- Flagship-class open weights from a European lab
- Priced well under the $2/$10 tier
- Vision built in
Cons
- Weights not yet downloadable at preview
- Independent index places it behind GLM-5.3, Kimi K3, and DeepSeek V4.1 Flash
- Context window figures disagree between Mistral and evaluators
Best for
Read more
Related comparisons
Alternatives to Mistral Large 4
DeepSeek V4.1 Flash
MIT-licensed open weights, native vision, a 1M context, and $0.15 per million input tokens off-peak.
Mistral Le Chat
Europe's answer to ChatGPT — fast, private, and GDPR-native.
Kimi
Moonshot AI's open-weight chat model, known for huge context windows and strong agentic benchmarks.