Pixtral is a VLM from Mistral AI (France).
One-line fit: best used for Document + image reasoning.
Key specifications
| Provider | Mistral AI |
|---|---|
| Model | Pixtral |
| Type | VLM |
| Context window | Large |
| Pricing (indicative) | API/open |
| Company category | Frontier Lab |
| Region | France |
| Official site | https://mistral.ai |
Benchmarks and public standing
Competitive open VLM
Treat scores as directional. SWE-bench, Terminal-Bench, LMSYS Arena, Artificial Analysis, and vendor cards use different harnesses and are not always comparable 1:1. Re-check the latest official model card before decisions.
What this model is good at
Document + image reasoning
This aligns with Mistral AI’s broader strengths:
- Open weights (Mixtral, Mistral)
- EU sovereign AI
- Coding models
- API simplicity
Limitations and watch-outs
Not video specialist
- Rate limits, regional availability, and data retention policies vary by plan.
- Agent harness quality (Cursor, Claude Code, Codex, custom tools) can change outcomes more than raw model Elo.
- Open-weight availability (if any) is separate from hosted API quality and safety filters.
Ideal users
- Product engineers shipping features that match: Document + image reasoning
- Teams standardizing on the Mistral AI ecosystem
- Agent builders who need a VLM
When to pick something else
- Cheaper volume: compare lower tiers from the same lab or open Chinese/EU alternatives.
- Maximum hard SWE thoughtfulness: compare Claude Fable-class and other coding flagships.
- Giant multimodal corpora / Workspace: compare Gemini-class models.
- Self-host / open weights: Llama, Qwen, DeepSeek, GLM, Mistral open lines.
Parent company
Mistral AI – Europe leading open-weight and commercial LLM lab; strong enterprise and sovereign-cloud story.
Related models from Mistral AI
- Mistral Large 2 / flagship class – Commercial flagship
- Mixtral 8x7B / 8x22B – Open MoE
- Codestral / Devstral – Code
Data is curated for AIForumSphere model directory mid-2026 directional figures.
