Models / MiMo V2.6 Flash
Best home rig for MiMo V2.6 Flash
Current model309B total · 15B active1 measured resultfastest plain decode: 35 tok/s model info ↗
Will it fit my GPU? Best value GPU per euro →
Real, measured generation speeds for MiMo V2.6 Flash on home hardware, fastest first. Different quants/engines aren't directly comparable.
| Build | Quant | Engine | tok/s | Prompt processing (tok/s) | Cost | Source |
|---|---|---|---|---|---|---|
| 2x DGX Spark: MiMo V2.6 Flash MXFP4 on SGLang, 35 tok/s (Mia's AI Lab recipe) | MXFP4 | SGLang, 2 nodes, 1 stream | 35 | – | github.com ↗ |
Prompt processing = prefill speed reported by the source, @ prompt length. * est. TTFT @8k is calculated from measured prefill, not measured. Why it matters.
Memory math: how much VRAM / RAM does MiMo V2.6 Flash need?
MiMo V2.6 Flash has 309B parameters, about 15B active per token (source: Xiaomi MiMo model card: sparse MoE, 309B total / 15B activated). Weight size = parameters × bits-per-weight ÷ 8, before context (KV cache) and runtime overhead.
| Format | Bits/weight | Weights | Smallest common memory size that holds the weights alone |
|---|---|---|---|
| Q8_0 | 8.50 | 328.3 GB | 512 GB |
| Q6_K | 6.56 | 253.5 GB | 256 GB |
| Q5_K_M | 5.70 | 220.3 GB | 256 GB |
| Q4_K_M | 4.89 | 189.0 GB | 192 GB |
| Q3_K_M | 4.00 | 154.3 GB | 192 GB |
| Q2_K | 3.16 | 122.0 GB | 128 GB |
Bits-per-weight for GGUF quants are llama.cpp's own figures (llama.cpp quantize README). "As measured" rows apply the source's stated bits-per-weight to the parameter count; real files differ slightly because some layers stay at higher precision. Try other combinations in the VRAM calculator.
Why it can still run on smaller GPUs: it is a mixture-of-experts model, so only about 15B of its 309B parameters are used for each generated token. Runtimes such as llama.cpp (--n-cpu-moe), ExLlamaV3 or Project Maya can keep most expert weights in system RAM and only the shared layers and hot experts in VRAM. Then you need VRAM + system RAM ≥ weights, and speed is limited mainly by RAM bandwidth.
FAQ
How much VRAM does MiMo V2.6 Flash need?
The weights are about 189.0 GB at Q4_K_M (309B params × 4.89 bits ÷ 8, before context/KV cache). Add memory for context (KV cache) and the runtime. The smallest-memory setup with a measured result here, fully in memory, is 2x DGX Spark (GB10) (256 GB) at 35 tok/s.
What is the fastest measured hardware for MiMo V2.6 Flash?
2x DGX Spark (GB10): 35 tok/s generation (SGLang, 2 nodes, 1 stream, MXFP4), from https://github.com/MiaAI-Lab/MiMo-V2.6-Flash-2x-DGX-Sparks.
Which GPUs have measured results for MiMo V2.6 Flash?
2x DGX Spark (GB10).
Can it run on… 2x DGX Spark (GB10)
Related: what GPU do you need for 2026 models · all builds