STACKQUADRANT

FareedKhan-dev/kimi-k3-in-c

Inference Engines

A 2.78-trillion-parameter Kimi K3 running inference on a single CPU in 8.24 GB of RAM. Portable C99: no BLAS, no framework, no GPU.

7.3
GitHub Metrics
Stars
6.6k
Forks
1.1k
Open Issues
5
Watchers
62
Contributors
10
Weekly Commits
13
Language
C
License
Apache-2.0
Last Commit
Aug 26, 2026
Created
Aug 1, 2026
Latest Release
v1.0.0
Release Date
Aug 7, 2026
Synced: Aug 28, 2026
Quality Scores
Documentation Qualityw: 20%
6.8

Has docs site (https://medium.com/@fareedkhandev/building-kimi-k3-in-c-to-run-a-2-8t-model-on-consumer-hardware-a5792cbf3b59). Description: 131 chars. Stars signal: 6,618. Contributors: 10. Score: 6.8/10

Community Healthw: 20%
6.6

Stars: 6,618. Contributors: 10. Watchers: 62. Forks: 1,079. Issue ratio: 0.1%. Score: 6.6/10

Maintenance Velocityw: 15%
9.4

Last commit: 2d ago. Weekly commits: 13. Latest release: v1.0.0. Score: 9.4/10

API Design & DXw: 20%
8.0

Stars/issues ratio: 1324. Has documentation site. Permissive license: Apache-2.0. Popularity signal: 6,618 stars. Score: 8/10

Production Readinessw: 15%
5.6

Battle-tested: 6,618 stars. Peer review: 10 contributors. Versioned: v1.0.0. Licensed: Apache-2.0. Age: 0.1 years. Maintenance: last commit 2d ago. Score: 5.6/10

Ecosystem Integrationw: 10%
7.7

Fork interest: 1,079. Ecosystem: C. Integration-friendly: Apache-2.0. Adoption: 6,618 stars. Has web presence. Score: 7.7/10

Tags
avx2c99cpu-inferencedeep-learningfrom-scratchinference-enginekimi-k3linear-attentionllmllm-inference
Radar
Documentation Quality
Community Health
Maintenance Velocity
API Design & DX
Production Readiness
Ecosystem Integration