Mistral Large 4
Mistral’s 1T-parameter multimodal MoE flagship (preview; weights promised).
Mistral Large 4 is a frontier multimodal (text and image input) model built for reasoning, coding and agentic workloads, with 1T total / 49B active parameters and a 1M-token context window. Open weights were promised but not yet released.
- Provider
- Mistral AI
- Type
- Language model
- Released
- Oct 6, 2026
- Context window
- 1M tokens
- Max output
- 262K tokens
- Price
- $0.68 input / $2.09 output per 1M tokens
- Input
- text, image
- Output
- text
- Open weights
- No
Best for
- Reasoning
- Coding
- Agents
- Enterprise
- Long context
Strengths
- European provider (data residency)
- Low price for its size
More from Mistral AI
- Mistral Medium 3.5 — Mistral AI. Fast enterprise-grade multimodal model.
- Mistral Small 4 — Mistral AI. Apache-licensed open model unifying reasoning, vision and coding.
- Voxtral Small — Mistral AI. Best open-weight transcription model — also understands audio.
Tools that use it
- Le Chat — Mistral AI. Mistral’s fast, European AI assistant.