ToolSite MiMo V2.6 Pro VRAM requirements
MiMo V2.6 Pro has 1.02T parameters, of which about 42B are used per token; all 384 experts still have to be in memory. With an 8K-token context and one request it needs about 636 GB of GPU memory at Q4_K_M, 1,050 GB at FP8 and 2,099 GB at FP16/BF16. The published weights take 527 GB (4.4 bits/weight). No single GPU in the table holds it at Q4_K_M; it needs a multi-GPU setup.
Open MiMo V2.6 Pro in the calculator
VRAM by quantization
Weights plus the KV cache for 8,192 tokens in FP16 and the runtime overhead (0.5 GB plus 10%). Each row opens the calculator with that setting. What the GGUF names mean.
| Precision | Weights | Total | Smallest setup |
|---|---|---|---|
| As published (4.4 bits/weight) | 527 GB | 581 GB | 4 × B200 |
| FP16 / BF16 | 1,908 GB | 2,099 GB | More than 8 GPUs |
| FP8 / INT8 | 954 GB | 1,050 GB | 8 × H200 |
| INT4 (AWQ / GPTQ) | 507 GB | 558 GB | 4 × H200 |
| GGUF Q8_0 | 1,013 GB | 1,116 GB | 8 × H200 |
| GGUF Q6_K | 782 GB | 861 GB | 8 × H200 |
| GGUF Q5_K_M | 676 GB | 745 GB | 8 × H200 |
| GGUF Q4_K_M | 577 GB | 636 GB | 4 × B200 |
| GGUF Q3_K_M | 466 GB | 514 GB | 4 × H200 |
| GGUF Q2_K | 399 GB | 440 GB | 4 × H200 |
KV cache at long context
10 of its 70 layers use full attention and 60 keep a sliding window of 128 tokens. Each extra token of context adds 50 KB of FP16 cache per request once the sliding windows are full. How the KV cache works.
| Context | KV cache, FP16 | KV cache, FP8 | Total at Q4_K_M |
|---|---|---|---|
| 4K tokens | 238 MB | 119 MB | 636 GB |
| 32K tokens | 1.60 GB | 819 MB | 637 GB |
| 128K tokens | 6.29 GB | 3.14 GB | 642 GB |
| 1M tokens | 50.0 GB | 25.0 GB | 690 GB |
Which GPUs can run MiMo V2.6 Pro
With 8,192 tokens of context. Several GPUs means one tensor-parallel group of 2, 4 or 8 cards.
| GPU | Memory | Q4_K_M | FP8 |
|---|---|---|---|
| RTX 3060 | 12 GB | Needs more than 8 | Needs more than 8 |
| RTX 4060 Ti 16GB | 16 GB | Needs more than 8 | Needs more than 8 |
| RTX 3090 / 4090 | 24 GB | Needs more than 8 | Needs more than 8 |
| RTX 5090 | 32 GB | Needs more than 8 | Needs more than 8 |
| A100 40GB | 40 GB | Needs more than 8 | Needs more than 8 |
| Mac, 64 GB unified memory about 75% of it is usable by the GPU by default | 48 GB | Needs more than 8 | Needs more than 8 |
| L40S / RTX 6000 Ada | 48 GB | Needs more than 8 | Needs more than 8 |
| A100 / H100 80GB | 80 GB | Needs 8 | Needs more than 8 |
| Mac, 128 GB unified memory about 75% of it is usable by the GPU by default | 96 GB | Needs 8 | Needs more than 8 |
| H200 | 141 GB | Needs 8 | Needs 8 |
| B200 | 180 GB | Needs 4 | Needs 8 |
How fast MiMo V2.6 Pro writes
Tokens per second for one request with 8,192 tokens of context, estimated from memory bandwidth and the parameters read per token. A dash means it does not fit on one card. Try other settings in the speed calculator.
| Hardware | Bandwidth | Q4_K_M | FP8 |
|---|---|---|---|
| RTX 3060 12GB | 360 GB/s | — | — |
| RTX 4090 | 1,008 GB/s | — | — |
| RTX 5090 | 1,792 GB/s | — | — |
| M4 Max Mac (128 GB) | 546 GB/s | — | — |
| M3 Ultra Mac Studio (512 GB) | 819 GB/s | — | — |
| H100 SXM | 3,350 GB/s | — | — |
| H200 | 4,800 GB/s | — | — |
Longest context on one GPU
How many tokens of context fit on a single card with one request and an FP16 KV cache. "Full" means the model's whole context window fits.
| GPU | Memory | Q4_K_M | Q8_0 | FP8 |
|---|---|---|---|---|
| RTX 3060 | 12 GB | No | No | No |
| RTX 4060 Ti 16GB | 16 GB | No | No | No |
| RTX 3090 / 4090 | 24 GB | No | No | No |
| RTX 5090 | 32 GB | No | No | No |
| A100 40GB | 40 GB | No | No | No |
| Mac, 64 GB unified memory | 48 GB | No | No | No |
| L40S / RTX 6000 Ada | 48 GB | No | No | No |
| A100 / H100 80GB | 80 GB | No | No | No |
| Mac, 128 GB unified memory | 96 GB | No | No | No |
| H200 | 141 GB | No | No | No |
| B200 | 180 GB | No | No | No |
Model details
- Parameters
- 1.02T (1,024,216,603,392)
- Experts
- 384 routed experts, all loaded
- Active per token
- 42B
- Layers
- 10 of its 70 layers use full attention and 60 keep a sliding window of 128 tokens
- Attention cache
- 8 KV heads × 192 keys, 128 values
- Sliding-window layers
- 8 KV heads × 192
- Context length
- 1,048,576 tokens
- Published weights
- 527 GB (4.4 bits/weight)
- On Hugging Face
- XiaomiMiMo/MiMo-V2.6-Pro-RL
Other models
- DeepSeek V4 Flash 0731
- Qwen3.6 35B-A3B
- Qwen3.6 27B
- Qwen3.8 27B
- Qwen3.8 Flash Next
- Qwen3.8 2.4T-A95B
- Qwen3.5 9B
- Qwen3.5 122B-A10B
- Qwen3-Coder-Next
- DeepSeek V4.1 Flash
- DeepSeek V4 Flash
- DeepSeek V4 Pro
- DeepSeek V3.2
- GLM-5.3
- GLM-5.3 Flash
- GLM-5.2
- GLM-4.7 Flash
- Gemma 4 31B
- Gemma 4 26B-A4B
- Gemma 4 12B
- Gemma 4 E4B
- Kimi K3
- MiniMax M3
- MiniMax M2.7
- MiMo V2.6 Flash
- Mistral Medium 3.5 128B
- Nemotron 3 Nano 4B
- Nemotron 3 Nano 30B-A3B
- Nemotron 3 Super 120B-A12B
- Xing 4.0 29B-A4B
- Muse Glimmer 30B
- MiniCPM5 2B
- Llama 3.1 8B
- Llama 3.1 70B
- Qwen3 8B
- Qwen3 30B-A3B
- gpt-oss-20b
- gpt-oss-120b
- DeepSeek V3 / R1
Numbers read from the model files on Hugging Face on .