1.FareedKhan-dev / kimi-k3-in-cA 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.avx2c99cpu-inferencedeep-learningfrom-scratchinference-engine+14 moreCAPACHE-2.010K total1.6K forks+7,410 starsfrom 2,865 · score 8.91
2.antirez / ds4DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCmCMIT24K total2.4K forks+3,596 starsfrom 20,636 · score 8.19
3.karpathy / llama2.cInference Llama 2 in one file of pure CCMIT20K total2.7K forks+371 starsfrom 19,881 · score 5.92
4.eunomia-bpf / agentsightlightweight system-level observability for AI AgentsagentebpfllmobservabilityCMIT771 total118 forks+220 starsfrom 551 · score 5.40