Instructions to use revorm/MiMo-V2.6-Flash-MOPD-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use revorm/MiMo-V2.6-Flash-MOPD-NVFP4 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="revorm/MiMo-V2.6-Flash-MOPD-NVFP4", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("revorm/MiMo-V2.6-Flash-MOPD-NVFP4", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use revorm/MiMo-V2.6-Flash-MOPD-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "revorm/MiMo-V2.6-Flash-MOPD-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "revorm/MiMo-V2.6-Flash-MOPD-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/revorm/MiMo-V2.6-Flash-MOPD-NVFP4

SGLang

How to use revorm/MiMo-V2.6-Flash-MOPD-NVFP4 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "revorm/MiMo-V2.6-Flash-MOPD-NVFP4" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "revorm/MiMo-V2.6-Flash-MOPD-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "revorm/MiMo-V2.6-Flash-MOPD-NVFP4" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "revorm/MiMo-V2.6-Flash-MOPD-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use revorm/MiMo-V2.6-Flash-MOPD-NVFP4 with Docker Model Runner:
```
docker model run hf.co/revorm/MiMo-V2.6-Flash-MOPD-NVFP4
```

MiMo-V2.6-Flash-MOPD-NVFP4

Lossless MXFP4 → NVFP4 transcode of XiaomiMiMo/MiMo-V2.6-Flash-MOPD (revision 2479e2d0029eca9a34cc7e7f55a121925f81908e). No calibration and no re-quantization: the dequantized expert weights are bit-identical to the official checkpoint.

What was changed

Routed experts (model.layers.*.mlp.experts.*.{gate,up,down}_proj): the FP4 (E2M1) values are copied bit-for-bit. Each 32-element E8M0 block scale becomes two 16-element E4M3 block scales of value 2^(e − 127 − k), with a per-tensor weight_scale_2 = 2^k and k = max block exponent − 8. gate_proj and up_proj of the same expert share k, because they are fused into one w13 matrix at load time. input_scale = 1.0.
Everything else is byte-identical to the official checkpoint: FP8 block-scaled linear layers (weight_scale_inv), BF16 tensors, MTP layers, vision and audio encoders, the DFlash draft model (dflash/), tokenizer and chat template.
Quantization config: ModelOpt MIXED_PRECISION (experts NVFP4, group size 16; other linear layers FP8_BLOCK_SCALES).

Verification

Tensor layout (145,273 tensors: names, dtypes, shapes) is identical to tiyuvta/MiMo-V2.6-Flash-RL-NVFP4, which uses the same transcode of the RL checkpoint (the same procedure reproduces its tensors byte-for-byte from the RL weights on all sampled tensors).
200 randomly sampled tensors are byte-identical to the independent transcode AxionML/MiMo-V2.6-Flash-MOPD-NVFP4.
SHA256SUMS lists every file.
Served with vLLM, tensor parallel 4: 17×23 = 391; 5/5 tool-calling checks (single tool, 8 tools, streaming, tool result, two calls per turn); needle retrieval at 16K / 40K / 114K tokens; GSM8K (first 100, temperature 0) 98/100.

Serving notes (vLLM)

Use --trust-remote-code --reasoning-parser mimo --tool-call-parser mimo --enable-auto-tool-choice.
Use --generation-config vllm: the shipped generation_config.json sets max_new_tokens: 2048, which truncates long reasoning answers. Recommended sampling: temperature=1.0, top_p=0.95.
Some vLLM builds route ModelOpt FP8_BLOCK_SCALES layers in mixed-precision checkpoints to the unquantized method, or fail with KeyError: ...qkv_proj.weight_scale_inv; these layers need vLLM's block-FP8 linear method.
Check the log line Using '...' NvFp4 MoE backend: FLASHINFER_CUTLASS runs native FP4; MARLIN is a W4A16 fallback (vLLM 0.30.x selects it whenever --enable-lora is set).
With the DFlash draft enabled, vLLM rejects min_p and logit_bias.

License

MIT, same as the base model.

Downloads last month: 79

Safetensors

Model size

159B params

Tensor type

F32

BF16

F8_E4M3

Model tree for revorm/MiMo-V2.6-Flash-MOPD-NVFP4

Base model

XiaomiMiMo/MiMo-V2.6-Flash-MOPD

Quantized

(22)

this model