What Do You Need to Run Mistral Locally? VRAM by Version
Mistral currently has 5 versions you can deploy locally. The smallest, Mistral 7B v0.3, needs 7GB of VRAM at Q4; the largest, Mistral Small 3.2 24B, needs 16GB. The cheapest device that gets one running is the Tesla V100 16GB, at about $131.
VRAM needed for each Mistral version
Xianyu
| Version | Released | Context | |||
|---|---|---|---|---|---|
| Mistral 7B v0.3Dense | 7.3B | May 22, 2024 | 32K | 7 | Tesla V100 16GB$131 · 16 GB |
| Ministral 3 8BDense | 8.9B | Oct 31, 2025 | 256K | 8 | Tesla V100 16GB$131 · 16 GB |
| Mistral NemoDense | 12.3B | Jul 17, 2024 | 128K | 10 | Tesla V100 16GB$131 · 16 GB |
| Ministral 3 14BDense | 14B | Oct 31, 2025 | 256K | 11 | Tesla V100 16GB$131 · 16 GB |
| Mistral Small 3.2 24BDense | 24B | Jun 19, 2025 | 128K | 16 | Tesla V100 16GB$131 · 16 GB |
Minimum VRAM = Q4_K_M weights + KV cache for an 8K context + 1GB runtime overhead, rounded up. The cheapest device is the first one that fits when sorting by median price on the current price basis.
About Mistral
Mistral is published by Mistral AI under Apache-2.0. This page covers 5 versions ranging from 7.3B to 24B parameters. New official releases are scanned weekly; architecture parameters and sourced or estimated weight sizes are reviewed before publication.
VendorMistral AI
LicenseApache-2.0
Official sitehttps://mistral.ai/
Hugging Face orgmistralai