Instructions to use patrickbdevaney/MiMo-V2.6-Flash-REAP50 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use patrickbdevaney/MiMo-V2.6-Flash-REAP50 with Transformers:

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

pipe = pipeline("text-generation", model="patrickbdevaney/MiMo-V2.6-Flash-REAP50", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("patrickbdevaney/MiMo-V2.6-Flash-REAP50", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use patrickbdevaney/MiMo-V2.6-Flash-REAP50 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "patrickbdevaney/MiMo-V2.6-Flash-REAP50"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "patrickbdevaney/MiMo-V2.6-Flash-REAP50",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/patrickbdevaney/MiMo-V2.6-Flash-REAP50

SGLang

How to use patrickbdevaney/MiMo-V2.6-Flash-REAP50 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 "patrickbdevaney/MiMo-V2.6-Flash-REAP50" \
    --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": "patrickbdevaney/MiMo-V2.6-Flash-REAP50",
		"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 "patrickbdevaney/MiMo-V2.6-Flash-REAP50" \
        --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": "patrickbdevaney/MiMo-V2.6-Flash-REAP50",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use patrickbdevaney/MiMo-V2.6-Flash-REAP50 with Docker Model Runner:
```
docker model run hf.co/patrickbdevaney/MiMo-V2.6-Flash-REAP50
```

MiMo-V2.6-Flash-REAP50

XiaomiMiMo/MiMo-V2.6-Flash-RL with 50% of its routed experts removed — 256 experts per layer down to 128 — so that it fits and serves on a single NVIDIA Jetson AGX Thor (117 GiB unified memory). 86.1 GiB across 65 shards.

Vision, audio and video input are preserved; audio_tokenizer/ ships with the checkpoint.

GGUF Quantizations (llama.cpp)

Official llama.cpp GGUF quantizations (including native MXFP4_MOE, optimal hybrid Q2_K, multimodal mmproj, and speculative mtp draft towers) are available at: 👉 patrickbdevaney/MiMo-V2.6-Flash-REAP50-GGUF

How the experts were chosen

Not by activation frequency. Expert saliency was accumulated over a calibration corpus and the prune set was solved with HOPE, which minimises the output error a prune set actually causes including the interaction terms between experts — REAP is the same objective with the off-diagonal zeroed, and that off-diagonal cannot be recovered after the pass.

setting	value
objective	`hope`
saliency criterion	`reap_1_1_1`
prune ratio	0.50, uniform across layers
per-domain protection	top 8% of every domain held out of the prune set
worst domain retained	0.9935 (audio)
mean retained	0.9964
HOPE objective pᵀFp	0.01624

Selection is scored per domain and ranked by the worst one, never the mean: an average is how a criterion that destroys one capability outscores one that preserves all of them.

Retained gated output mass, by calibration domain

domain	retention
audio	0.9935
image	0.9937
video	0.9938
science	0.9973
math	0.9974
finance	0.9974
ballast	0.9976
code	0.9983
agentic	0.9983

Routers

Pruning an expert leaves its router column behind. The routers were refitted by output matching against the unpruned teacher, routers only, every expert frozen: 47 routers refitted; all 47 kept at the teacher weights (the fit did not beat the baseline).

A refit that failed to beat the untouched baseline was discarded in favour of the baseline, so no router here is worse than simply slicing the teacher's.

Limitations

Calibration was English/Chinese text, code, math, science, finance, agentic traces, and image/audio/video captions. Domains outside that mix were not measured.
The dflash/ speculative-decoding draft head from the source repo is not included: it was trained against the unpruned expert set and is not valid for this checkpoint.
Pruned MoE experts do not come back. This is a lossy, irreversible transform of the base model.

Provenance

Produced by patrickbdevaney/xiaomi-2.6-flash-REAP on a single Jetson AGX Thor. MIT, inherited from the base model — attribution to Xiaomi MiMo.

Downloads last month: 420

Safetensors

Model size

159B params

Tensor type

BF16

F8_E4M3

F32

Model tree for patrickbdevaney/MiMo-V2.6-Flash-REAP50

Base model

XiaomiMiMo/MiMo-V2.6-Flash-RL

Quantized

(34)

this model

Quantizations

1 model