# MoonshotAI/Kimi-K2

Kimi K2 is the large language model series developed by Moonshot AI team

Repository: https://github.com/MoonshotAI/Kimi-K2
Canonical: https://ross.abutalabs.com/products/kimi-k2
License: NOASSERTION
License Family: other
Last push: 2026-01-21T05:33:12+00:00

## Health v2 (maintenance only)
Score: 47/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 63, release rhythm 35, longevity 30
- inputs: {"age_days": 426, "days_push": 224, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 11105, forks 908 (observed 2026-08-28T04:10:46.119063+00:00)

## What it is
Kimi K2 is Moonshot AI's open-weight mixture-of-experts large language model series with 1 trillion total parameters (32B activated), optimized for agentic tool use, reasoning, and coding. The repository hosts model documentation, weights links, and inference examples for the Base and Instruct variants.

## Use cases
- download and run an open-weight trillion-parameter MoE LLM
- build autonomous agents with strong tool-use capabilities
- fine-tune a foundation model for custom applications
- serve a chat model for general-purpose assistant experiences
- research large-scale MoE training with the Muon optimizer

## When to choose
- you need frontier-level open weights for agentic or coding tasks
- you have GPU infrastructure to serve a 1T-parameter MoE model
- you want a base model for full fine-tuning control

## When to avoid
- you lack the hardware to host a trillion-parameter model
- you need a small model for local or edge deployment
- you need a permissively licensed model without restrictions (license is modified MIT)

## Facets
- artifact type: learning-resource
- maturity: active
- function: llm-inference, llm-training, agent-framework, machine-learning
- domain: large-language-models, artificial-intelligence, deep-learning
- platform: python, cloud
- tags: mixture-of-experts, open-weights, moonshot-ai, agentic-intelligence, model-weights, huggingface, ai-agents, gpu

## Member repositories
- MoonshotAI/Kimi-K2 (main) score 47

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:46.119063+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 2026-08-29T17:16:47.755169+00:00, confidence not recorded.
  - readme: https://github.com/MoonshotAI/Kimi-K2 (fetched 2026-08-28T04:10:46.119063+00:00, sha 194c78639254)
- Data as of 2026-08-30T08:39:29.467469+00:00.