# MoonshotAI/Kimi-K3

Open Frontier Intelligence

Repository: https://github.com/MoonshotAI/Kimi-K3
Canonical: https://ross.abutalabs.com/products/kimi-k3
License: NOASSERTION
License Family: other
Last push: 2026-08-06T10:17:28+00:00

## Health v2 (maintenance only)
Score: 56/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 96, release rhythm 35, longevity 2
- inputs: {"age_days": 37, "days_push": 27, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, young, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 8632, forks 700 (observed 2026-08-28T04:10:23.912707+00:00)

## What it is
Kimi K3 is an open-weight, 2.8T-parameter native multimodal agentic model from Moonshot AI, built on Kimi Delta Attention and Attention Residuals with a 1M-token context window. The repository hosts model weights, license, and technical report for the first open 3T-class frontier model.

## Use cases
- run a frontier open-weight LLM locally
- long-horizon autonomous coding agent
- deep research with visualizations
- multimodal reasoning over long documents
- agentic knowledge work and tool orchestration

## When to choose
- you need an open-weight frontier model with 1M-token context
- you want native multimodal agentic capabilities you can self-host
- you need long-horizon coding or research automation

## When to avoid
- you lack large-scale GPU infrastructure for a 2.8T-parameter MoE model
- you need a permissively licensed model (custom Kimi K3 license)
- you only need a small or lightweight model

## Facets
- artifact type: learning-resource
- maturity: active
- function: llm-inference, machine-learning, agent-framework, rag
- domain: large-language-models, artificial-intelligence, deep-learning
- platform: python, self-hosted
- tags: open-weights, multimodal, mixture-of-experts, long-context, frontier-model, moonshot-ai, kimi, ai-agents, gpu

## Member repositories
- MoonshotAI/Kimi-K3 (main) score 56

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:23.912707+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 2026-08-29T17:25:51.818170+00:00, confidence not recorded.
  - readme: https://github.com/MoonshotAI/Kimi-K3 (fetched 2026-08-28T04:10:23.912707+00:00, sha 1fe322c0d7d5)
- Data as of 2026-08-30T08:39:29.467469+00:00.