Can MiMo 7B RL run on the H100 SXM?
Yes. At 8K context the best fit is F16, using about 16.4 GB of the 76.0 GB usable on the H100 SXM. Estimated speed ceiling: ~214 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|
| 2-bit (Q2_K) | 4.0 GB Fits | 4.6 GB Fits | 7.9 GB Fits | 1140 |
| 3-bit (Q3_K_M) | 4.8 GB Fits | 5.4 GB Fits | 8.8 GB Fits | 875 |
| 4-bit (Q4_K_M) • | 5.6 GB Fits | 6.2 GB Fits | 9.6 GB Fits | 715 |
| 5-bit (Q5_K_M) | 6.4 GB Fits | 7.0 GB Fits | 10.4 GB Fits | 600 |
| 6-bit (Q6_K) • | 7.1 GB Fits | 7.6 GB Fits | 11.0 GB Fits | 535 |
| 8-bit (Q8_0) • | 8.8 GB Fits | 9.4 GB Fits | 12.7 GB Fits | 413 |
| 16-bit (F16/BF16) | 15.8 GB Fits | 16.4 GB Fits | 19.8 GB Fits | 214 |
• real GGUF file size from jedisct1/MiMo-7B-RL-GGUF. Speed is a bandwidth ceiling (3350 GB/s), not a benchmark. Methodology.
More
MiMo 7B RL on every GPU and Mac · Every model that runs on the H100 SXM · Other models that fit 64 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (XiaomiMiMo/MiMo-7B-RL), NVIDIA specs. Data as of 2026-11-17.