Mistral on NVIDIA Blackwell

Can I run Mistral 7B v0.3 on an NVIDIA DGX Spark (GB10)?

Yes. Mistral 7B v0.3 at Q4_K_M uses 5.6 GB of the 96.0 GB available on a NVIDIA DGX Spark (GB10), and runs at about 38.8 tokens per second. You can push the context to 32K.

Runs comfortably

5.6 GB of 96.0 GB · 6%
096 GB
Weights 4.1 GB
KV cache 1.0 GB
Runtime overhead 0.5 GB

Mistral 7B v0.3 at Q4_K_M leaves 90.4 GB spare on a NVIDIA DGX Spark (GB10). There is room to raise the context length or move up a quantisation level.

Generation38.8tok/s
Prompt processing3621tok/s
Max context32Ktokens
KV per 1K tokens0GB

Every quantisation of Mistral 7B v0.3 on a NVIDIA DGX Spark (GB10)

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 13.5 GB 15.0 GB Runs comfortably 32K 12.9
INT8 / W8A8 8.50 7.1 GB 8.6 GB Runs comfortably 32K 23.6
Q8_0 (GGUF) 8.50 7.1 GB 8.6 GB Runs comfortably 32K 23.6
FP8 (E4M3) 8.00 6.8 GB 8.3 GB Runs comfortably 32K 24.8
Q6_K 6.56 5.5 GB 7.1 GB Runs comfortably 32K 29.7
Q5_K_M 5.67 4.8 GB 6.3 GB Runs comfortably 32K 33.9
Q5_K_S 5.52 4.7 GB 6.2 GB Runs comfortably 32K 34.7
Q4_K_M 4.85 4.1 GB 5.6 GB Runs comfortably 32K 38.8
AWQ 4-bit 4.25 4.0 GB 5.5 GB Runs comfortably 32K 40.2
GPTQ 4-bit 4.25 4.0 GB 5.5 GB Runs comfortably 32K 40.2
MXFP4 4.25 4.0 GB 5.5 GB Runs comfortably 32K 40.2
Q4_K_S 4.58 3.9 GB 5.4 GB Runs comfortably 32K 40.7
Q4_0 4.55 3.9 GB 5.4 GB Runs comfortably 32K 41.0
IQ4_XS 4.25 3.6 GB 5.1 GB Runs comfortably 32K 43.4
Q3_K_M 3.91 3.3 GB 4.9 GB Runs comfortably 32K 46.5
IQ3_M 3.70 3.2 GB 4.7 GB Runs comfortably 32K 48.6
IQ3_XXS 3.06 2.7 GB 4.2 GB Runs comfortably 32K 56.5
Q2_K 2.63 2.3 GB 3.8 GB Runs comfortably 32K 63.4
IQ2_XXS 2.06 1.8 GB 3.4 GB Runs comfortably 32K 75.8
IQ1_M 1.75 1.6 GB 3.1 GB Runs comfortably 32K 84.8

Also worth checking

Mistral 7B v0.3 on every GPU
Full hardware tableEverything that runs on a NVIDIA DGX Spark (GB10)
Full model tableWhy the KV cache is the part that bites
Guide