QriticalBeta
Open Model Catalogue
Mistral AI
Mistral AI
Text generationVision

Mistral Small 4

mistralai/Mistral-Small-4-119B-2603

Mistral open MoE — Apache 2.0, 2 × 80 GB

Deploy Mistral Small 4 nowView on HuggingFace
Parameters
119B MoE · FP8 · vision
Min VRAM
≈ 150 GB
Hardware
2 × 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.