# XiaomiMiMo/MiMo

MiMo: Unlocking the Reasoning Potential of Language Model – From Pretraining to Posttraining

Repository: https://github.com/XiaomiMiMo/MiMo
Canonical: https://ross.abutalabs.com/products/mimo
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2025-06-05T16:01:49+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 25, release rhythm 35, longevity 35
- inputs: {"age_days": 494, "days_push": 454, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2299, forks 105 (observed 2026-08-28T04:06:35.330203+00:00)

## What it is
Xiaomi's MiMo is a 7B-parameter reasoning language model trained from pretraining through posttraining with reinforcement learning, released with model weights on HuggingFace and a technical report. The repository provides the training methodology, model checkpoints, and evaluation results showing strong math and code reasoning performance.

## Use cases
- run a small reasoning LLM for math problems
- evaluate a 7B model on AIME and LiveCodeBench
- study RL posttraining for language model reasoning
- download open weights for a reasoning model
- compare small reasoning models against DeepSeek R1
- fine-tune a base model for code and math reasoning

## When to choose
- you need an open-weights 7B reasoning model for math or code
- you want to reproduce or study pretraining-to-posttraining RL pipelines
- you need strong reasoning performance on limited hardware

## When to avoid
- you need a general-purpose chat assistant rather than a reasoning model
- you require a model larger than 7B parameters
- you want a production inference server rather than model weights and training research

## Facets
- artifact type: library
- maturity: active
- function: llm-training, llm-inference, machine-learning, deep-learning
- domain: large-language-models, machine-learning, deep-learning, artificial-intelligence
- platform: python
- tags: reasoning-model, reinforcement-learning, pretraining, posttraining, math-reasoning, code-reasoning, 7b-model, xiaomi, gpu, linux

## Member repositories
- XiaomiMiMo/MiMo (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:35.330203+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:40:15.983509+00:00, confidence not recorded.
  - readme: https://github.com/XiaomiMiMo/MiMo (fetched 2026-08-28T04:06:35.330203+00:00, sha 8d3e18bfe997)
- Data as of 2026-08-30T08:39:29.467469+00:00.