QriticalBeta
Open Model Catalogue
Mistral AI
Mistral AI
Text generationVision

Ministral 3 8B

mistralai/Ministral-3-8B-Instruct-2512

Mistral small — Apache 2.0, vision, any GPU

Deploy Ministral 3 8B nowView on HuggingFace
Parameters
8B · FP8 · vision
Min VRAM
≈ 20 GB
Hardware
1 × 80 GB GPU
OpenAI-compatible API with a per-workload key — drop-in for existing SDKs.
Dedicated GPUs — your model, your instance, no shared queues.
Billed by the second with auto-shutdown — never pay for idle time you didn't plan.