kimi-k3-in-c

kimi-k3-in-c is a local inference server - 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.

Overview

Kimi K3 inference in portable C99. No BLAS. No framework. No GPU.

2.78T parameters 1.56 TB checkpoint on disk 8.24 GB peak RSS, measured 176 KB the whole engine 0 GPUs

The same 2.78-trillion-parameter model, the same answer, on whatever machine you own. More memory only buys speed:

an ordinary laptop 8 GB 26.5 s the whole model streams off the disk on every step

a high-end laptop 32 GB 24.2 s some of the model now sits in memory

a heavy workstation 128 GB+ 5.6 s the model fits entirely in memory, the disk wait is gone

kimi-k3-in-c is open-source, written primarily in C, with 5,613 GitHub stars under the Apache 2.0 license. The latest release is v1.0.0 (2026-08-07).

Key capabilities

From the project's documentation:

• Quick start: clone, build and verify in about a minute, with no model
• Full setup: the whole path to generated text
• Reading the run report
• The problem: a model that does not fit
• The machine, and what it assumes
• 1. Reading a 1.56 TB checkpoint from its headers

How it fits a local-AI stack

kimi-k3-in-c runs on your own hardware, so pair it with a model and a GPU sized to your needs. Use the VRAM calculator to pick a model that fits your card, and see what you can run for hardware guidance. Related local inference servers in the directory:

• "BentoML"
• "candle-vllm"
• ExLlamaV2
• "flashinfer"

Sources

• Source code & docs: FareedKhan-dev/kimi-k3-in-c
• Official website: https://medium.com/@fareedkhandev/building-kimi-k3-in-c-to-run-a-2-8t-model-on-consumer-hardware-a5792cbf3b59

Stats from GitHub, 2026-08-15.