MiMo-7B-RL

Name: MiMo-7B-RL
Author: Xiaomi (MiMo)

Xiaomi (MiMo) · MiMo · released 2025-04-30 · data checked 2026-11-01 Needs verification

MiMo-7B-RL is a 7.8B-parameter dense model from Xiaomi (MiMo) (China), released 2025-04-30 under the MIT license. It scores 40.7 on the Small LM Index, #18 of 26 small models. It needs about 6.6 GB of memory at 4-bit with an 8K-token context.

Small LM Index

40.7

#18 of 26 scored small models

Likely 36.1–45.3 · from 4 of 6 core benchmarks

Parameters

7.8B

Dense

Price per 1M tokens

—

No hosted API on OpenRouter; self-host

Context window

32K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best small model on that benchmark. Sources: 4 reported by Xiaomi (MiMo). How we label sources.

MMLU-Pro Lab-reported Knowledge58.6

Class best: Qwen3.5 9B 82.5

GPQA Diamond Lab-reported Knowledge54.4

Class best: Qwen3.5 9B 81.7

AIME Lab-reported Math · AIME 202555.4

Class best: Apriel 1.6 15B Thinker 88.0

LiveCodeBench Lab-reported Coding · LiveCodeBench v649.3

Class best: Granite 4.2 8B 73.2

Not reported: estimated

What the index model expects from this model's other scores. Not a measurement.

Humanity's Last Exam≈ 4

SWE-bench Verified≈ 21

Specifications

Lab: Xiaomi (MiMo) (CN · China)
Architecture: Dense transformer
Total parameters: 7.8B
Active parameters: All (dense)
Size class: Small
Context window: 32K tokens
Input modalities: text
Reasoning: Reasoning (thinking)
License: MIT
Openness: Open weights
Memory to run: ~6.6 GB at 4-bit, 8K context (more sizes)
Hugging Face: 31.9K downloads (30 days) · 281 likes

Weights on Hugging Face ↗

Compare with

MiMo-7B-RL vs Falcon H1R 7B (7.6B, index 57.5)
MiMo-7B-RL vs Hunyuan 7B Instruct (7.5B, index 52.5)
MiMo-7B-RL vs OLMo 3 7B Think (7.3B, index 43.4)
MiMo-7B-RL vs Gemma 3n E4B (7.8B, index 18.2)
MiMo-7B-RL vs Ling 3.0 Tiny (7.9B, index 51.9)
MiMo-7B-RL vs Edge0 8B-A1B (preview) (7.9B, index 47.2)

Sources

https://huggingface.co/XiaomiMiMo/MiMo-7B-RL

Something wrong here? Report it.

Run it locally

No GGUF build from a well-known publisher yet.

MLX (Mac)

pip install -U mlx-lm
mlx_lm.chat --model mlx-community/MiMo-7B-RL-4bit

vLLM (GPU server)

vllm serve XiaomiMiMo/MiMo-7B-RL

Full-precision weights, for a server GPU.

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context
16-bit F16	17 GB	21 GB
8-bit Q8_0	9.5 GB	14 GB
4-bit Q4_K_M	6.0 GB	10 GB

Check it against your GPU, Mac or phone →