Mistral-Nemo has 12.2B parameters. In BF16 its weights alone take 22.8 GiB; quantized to 4-bit, about 7.02 GiB. Each 32K-token request adds 5.00 GiB of KV cache. Cheapest way to run it today: 1 × GeForce RTX 3090 (24 GB) at $0.143 per hour.
Read every day from Hugging Face. Split files are added together; vision projectors are left out.
How many GPUs to run Mistral-Nemo
Fewest GPUs that fit the weights plus one 8K-token request. Data-center GPUs use vLLM (90% of memory usable), the others llama.cpp. Several GPUs split the model with tensor or pipeline parallelism.
GPU
BF16 / FP16
8-bit
4-bit
GeForce RTX 4090 24 GB
2
1
1
GeForce RTX 5090 32 GB
1
1
1
RTX PRO 6000 Blackwell 96 GB
1
1
1
Mac M5 Max 128 GB 96 GB usable
1
1
1
NVIDIA H100 80 GB
1
1
1
NVIDIA H200 141 GB
1
1
1
NVIDIA B200 (HGX) 180 GB
1
1
1
8-bit: FP8 with vLLM (INT8 on GPUs without FP8), Q8_0 with llama.cpp. 4-bit: AWQ / GPTQ with vLLM, Q4_K_M with llama.cpp, both counted at the Q4_K_M size.
Cheapest way to run Mistral-Nemo today
For each GPU: the fewest cards that fit Mistral-Nemo with one 8K-token request, the most faithful weight format at that count, and the cheapest on-demand price (2026-10-11).
On-demand prices from RunPod, Vast.ai, Verda and Azure, checked every hour. Serving many users needs more KV cache, so more memory: size it in the calculator. Some links are affiliate links: StudioTV may earn a commission, at no extra cost to you.
KV cache: how memory grows with context
Architecture: 40 layers · 40 full attention (8 KV × 128). Every layer keeps keys and values for every token (grouped-query attention), so the cache grows linearly with context.
Context per request
BF16 cache
FP8 cache
8K tokens
1.25 GiB
0.63 GiB
32K tokens
5.00 GiB
2.50 GiB
128K tokens
20.0 GiB
10.0 GiB
Mistral-Nemo VRAM FAQ
How much VRAM does Mistral-Nemo need?
In BF16 the weights alone take 22.8 GiB (24 GB). In FP8 that is 12.7 GiB, and about 7.02 GiB with 4-bit quantization (Q4_K_M). Each request then adds KV cache: 1.25 GiB at 8K tokens and 5.00 GiB at 32K (BF16 cache).
Can Mistral-Nemo run on a single RTX 4090 (24 GB)?
Yes. In Q8_0 it fits on one RTX 4090 with an 8K-token context (13.9 GiB including the cache, with llama.cpp).
What is the cheapest way to run Mistral-Nemo?
On 2026-10-11, the cheapest on-demand setup is 1 × GeForce RTX 3090 (24 GB) with Q8_0 weights on Vast.ai, at $0.143 per hour (about $104.39 per month). Next: 1 × NVIDIA L4 (24 GB) with FP8 weights on Vast.ai, at $0.268 per hour. Sized for one 8K-token request; prices are checked every hour.
How many H100 GPUs does Mistral-Nemo need?
With one request and an 8K-token context: 1 in BF16, 1 in FP8 and 1 in 4-bit (vLLM, 90% of memory usable). Serving many users at once needs more memory for their KV caches: the calculator sizes that for you.
Mistral-Nemo VRAM badge
For a model card or a README: the 4-bit size and the smallest GPU it fits on, linked to this page. Data as JSON: /api/models/mistral-nemo.json.
Estimates, not guarantees: computed from the official config.json with the same engine as the LLM VRAM Calculator. Real usage depends on your engine version and settings.
Mistral AI, France · Released July 17, 2024