Ministral 3 8B
Mistral AI · Providers EU
Sovereignty and hosting
This model is catalogued for reference. It is not hosted on LLM EU yet, and no first-party endpoint serves it here. Requests to it go to the provider directly, under that provider's own terms.
Summary
Ministral 3 8B is a small Apache-2.0 instruct model with vision input and a 256k-token context window. Mistral targets edge deployment with it: the vendor states the FP8 checkpoint fits in 12GB of VRAM and less when further quantised. Published self-reported scores put it close to Qwen3-VL-8B-Thinking on AIME and GPQA.
Specifications
- Provider
- Mistral AI (FR)
- Context window
- 262k
- Parameters
- 8.92B
- Modalities
- text, vision
- License
- apache-2.0 · licence
- Open weights
- yes
- Release date
- 31 Oct 2025
- Tasks
- cheap, vision, reasoning, summarization
Languages
Strong: en, fr, es, de, it, pt
Adequate: nl, zh, ja, ko, ar
Languages
Strong
- en
- fr
- es
- de
- it
- pt
Adequate
- nl
- zh
- ja
- ko
- ar
Not verified: the provider has not documented this claim. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.
When to use it
Strengths
- runs on a single small accelerator or edge device
- vision input at the 8B size class
- long 256k-token context window for its size
- Apache-2.0 licence
Limits
- benchmark figures are self-reported by the vendor
- small size caps headroom on hard reasoning and long agentic tasks
- no evaluation has been run by LLM EU