Xiaomi MiMo - Mixture of experts - verified
Run MiMo-V2.6-Flash-RL locally
MiMo-V2.6-Flash-RL is a 310.8B mixture-of-experts (15B active) model from Xiaomi MiMo with a 1.05M-token context window, released under MIT (commercial use is allowed); the smallest listed download is 167.4 GB, the MXFP4 GGUF build from ggml-org/MiMo-V2.6-Flash-RL-GGUF.
Official weights: XiaomiMiMo/MiMo-V2.6-Flash-RL - modalities: text, image, video, audio
- Total params
- 310.8B
- Active params
- 15B
- Context
- 1.05M
- Smallest download
- 167.4 GB
Fits in
Smallest download at most 85% of memoryDownloads
The official weights and every quantization listed for this model, with file sizes from Hugging Face.
| Download | Format | Size | Repo |
|---|---|---|---|
Official weights | MXFP4 | 172.9 GB | XiaomiMiMo/MiMo-V2.6-Flash-RL |
MXFP4Smallest listed | GGUF | 167.4 GB | ggml-org/MiMo-V2.6-Flash-RL-GGUF |
GB is decimal: 1 GB = 10^9 bytes. Add memory for the context and the operating system on top of the file size.
Run it
Copied verbatim from the page linked under each command. Server commands from model cards often assume a multi-GPU node; check the flags before you run them.
SGLang
sglang serve \
--trust-remote-code \
--model-path XiaomiMiMo/MiMo-V2.6-Flash-RL \
--tp 8 \
--dp 2 \
--enable-dp-attention \
--enable-dp-lm-head \
--mm-enable-dp-encoder \
--mem-fraction-static 0.65 \
--chunked-prefill-size 16384 \
--speculative-algorithm EAGLE \
--speculative-num-steps 3 \
--speculative-eagle-topk 1 \
--speculative-num-draft-tokens 4 \
--enable-multi-layer-eagle \
--reasoning-parser mimo \
--tool-call-parser mimo \
--host 0.0.0.0 \
--port 30000vLLM
vllm serve XiaomiMiMo/MiMo-V2.6-Flash-RL \
--tensor-parallel-size 4 \
--trust-remote-code \
--gpu-memory-utilization 0.95 \
--max-model-len auto \
--reasoning-parser mimo \
--tool-call-parser mimo \
--enable-auto-tool-choice \
--generation-config vllmNot sure which runtime to use? Compare local LLM runtimes by platform, API and features.
License
- License
- MIT
mit - Commercial use
- Commercial use: YesCommercial use is allowed.
- Gated download
- No. The files download directly.
Worth knowing
The Flash variant is the smaller MIT-licensed MiMo-V2.6 model, served with four-way tensor parallelism in the official vLLM command.
Released (per the announcement).
FAQ
How big is the MiMo-V2.6-Flash-RL download?
The official MXFP4 weights in XiaomiMiMo/MiMo-V2.6-Flash-RL are 172.9 GB. The smallest listed download is 167.4 GB, the MXFP4 GGUF build from ggml-org/MiMo-V2.6-Flash-RL-GGUF. Plan for more memory than the file size: the context (KV cache) and the operating system need room too.
Can I use MiMo-V2.6-Flash-RL commercially?
MiMo-V2.6-Flash-RL is released under MIT, and commercial use is allowed. The license text is at https://opensource.org/licenses/MIT.
How much memory do I need for MiMo-V2.6-Flash-RL?
By this directory's rule of thumb it fits in 256 GB and up: the smallest listed download, 167.4 GB, is at most 85% of 256 GB, which leaves room for the context and the operating system.
What is the context length of MiMo-V2.6-Flash-RL?
1,048,576 tokens. A longer context needs more memory for the KV cache, so you can run with a shorter one when memory is tight.
Sources
Last verified from the pages below.
- Hugging Face model card - huggingface.co
- HF API - huggingface.co
- config.json - huggingface.co
- MiMo-V2.6 announcement - mimo.xiaomi.com
Newsletter
New open weights, every week
One email per week. New model releases, open-weight drops and what they mean for what you build.
Keep exploring
Best local AI models
Our curated picks by hardware tier, with benchmarks and the reasoning.
Model API pricing and specs
First-party prices per 1M tokens when you would rather call an API.
AI benchmark leaderboards
Sourced scores, for when you need to compare quality.
Setup guides
Point your coding tools at the model you picked.