Mistral Large 4
Mistral's 'Le Chonk': a 1.05T-total / 52B-active natively multimodal mixture-of-experts model with a 1M-token context window and low / medium / high reasoning modes (high default). Public API preview on 6 Oct 2026, open weights on 29 Oct 2026 (the release date shown; the docs list general availability on 5 Nov). $0.70/$2.10 per 1M tokens with cached input at $0.07, roughly half the preview rate. Weights on Hugging Face as Mistral-Large-4-1T-A52B under the Mistral Ecosystem Open-Weight license, a modified MIT: internal use, fine-tuning and products built on it are fine, but a company offering it as a hosted model service needs a separate commercial licence (Large 3 was Apache 2.0). Mistral's own numbers (vendor-reported): Terminal-Bench 4.0 41%, FrontierCode 2.0 19.7%, BrowseComp 84.5%, and it wins 64% of blind Surge Code comparisons against GLM-5.3 and Kimi K3 but only 22% against Claude Opus 5.5. Artificial Analysis Intelligence Index v5.0.1 scores it 26 at high effort at $2.06 per Index task and 120 tokens/s — a coding-agent model first, well below Kimi K3.1 Pro (46) on general intelligence. Maximum output is not published; the figure shown is a placeholder.
1M
tokens
33K
tokens
$0.7
per 1M tokens
$2.1
per 1M tokens
120
tokens/sec
Oct 2026
2026-10-29
$1.40
per 1M tokens
62.1
quality per $
Capabilities
Benchmarks
Price vs Quality
Context Window in Context
- ▸ Mistral Large 41M
- GPT-4.11M
- Gemini 2.5 Pro1M
- Gemini 2.0 Flash1M
- Llama 4 Maverick1M
- GPT-5.51M
- GPT-5.5 Pro1M
- Claude Opus 4.71M
Compare With
Open Source: Licensed under Mistral Ecosystem Open-Weight License (modified MIT)
Noch nicht auf unserem Gateway
Mistral Large 4 ist open-weight, du kannst es also selbst betreiben. Swfte Connect macht aus deinem Deployment einen verwalteten Endpoint, der dieselbe API spricht wie jedes andere Modell in diesem Verzeichnis.
curl https://api.swfte.com/agents/v2/gateway/chat/completions \
-H "Authorization: Bearer $SWFTE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "mistral:mistral-large-4",
"messages": [{"role": "user", "content": "Hello"}]
}'About Mistral Large 4
Mistral Large 4 is a open-source AI model by Mistral AI, released on October 29, 2026. It supports a context window of 1000K tokens and can generate up to 33K output tokens.
At $0.7 per million input tokens and $2.1 per million output tokens, its blended cost of $1.40/1M tokens puts it in the mid-range pricing tier. Its value score of 62.1 reflects the balance of quality and cost.
Mistral Large 4 is available as an open-source model under the Mistral Ecosystem Open-Weight License (modified MIT) license, meaning you can self-host it for predictable costs or use it through API providers like Swfte Connect.
Using Mistral Large 4 with Swfte
Access Mistral Large 4 through Swfte Connect, our unified LLM gateway. Connect gives you a single API for 50+ models, with automatic routing, cost optimization, and fallback handling. You can also try Mistral Large 4 in our AI Playground before integrating.