Mistral AI
Mistral Large 3 (200B)
Mistral's open-weight flagship for high-quality enterprise use cases
Model Summary
Family
Mistral
Version
3.0
Parameters
200B (est.)
Parameter counts for closed models are estimates; vendors rarely publish exact sizes.
VRAM Requirements by Quantization
Memory needed to serve Mistral Large 3 (200B) for inference, including a 20% overhead for activations and KV cache.
| Precision | VRAM needed | Smallest single GPU that fits |
|---|---|---|
| INT4 (4-bit) | 111.76 GB | NVIDIA H200 SXM5 (141 GB) |
| INT8 (8-bit) | 223.52 GB | AMD Instinct MI325X (256 GB) |
| FP16 (16-bit) | 447.03 GB | Multi-GPU required |
| FP32 (32-bit) | 894.07 GB | Multi-GPU required |
Recommended GPU Configurations
Cheapest on-demand configurations to serve Mistral Large 3 (200B) at 8-bit (224 GB VRAM).
1x AMD Instinct MI325X
256 GB total VRAM · CDNA 3
~$8.00/h
2x AMD Instinct MI300X
384 GB total VRAM · CDNA 3
~$12.00/h
1x AMD Instinct MI355X
288 GB total VRAM · CDNA 4
~$13.00/h
Quick GPU Planning
Use the calculator pre-filled with this exact version to estimate memory, speed, and compute requirements in a few clicks.
Access Pre-filled CalculatorFrequently Asked Questions
How much VRAM do you need to run Mistral Large 3 (200B)?
With an estimated 200B parameters, Mistral Large 3 (200B) needs roughly 224 GB of VRAM in 8-bit (INT8), 112 GB in 4-bit, and 447 GB in FP16, including a 20% overhead for activations and KV cache.
Which GPUs can run Mistral Large 3 (200B)?
At 8-bit quantization, the most cost-effective option is 1x AMD Instinct MI325X (256 GB combined VRAM, around $8.00/hour on-demand). Higher-end cards like the NVIDIA B200 or AMD MI355X reduce the GPU count needed.
Can Mistral Large 3 (200B) run on a single GPU?
Yes. In 8-bit, a single AMD Instinct MI325X (256 GB) fits the model.
How much does it cost to serve Mistral Large 3 (200B) in the cloud?
Renting 1x Instinct MI325X costs on the order of $8.00/hour, i.e. about $5,840/month running 24/7. Actual prices vary by provider and commitment; spot and reserved capacity can be significantly cheaper.
Deploy on a GPU cloud
Rent 1x Instinct MI325X by the hour instead of buying hardware.
