ai-agent

kimi-k3-in-c

Name: kimi-k3-in-c
Author: FareedKhan-dev

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.

9,178 stars+96★ this week1,500 forksCUpdated 10/2/2026100% free · open source

What it does

Kimi-k3-in-c is a lightweight, 2.78-trillion-parameter AI model that runs on a single CPU, allowing for portable and efficient inference without relying on GPUs or external frameworks.

When to use it

•When you need to deploy a large language model on a resource-constrained device or environment
•When you want to avoid dependencies on external frameworks or libraries like BLAS
•When you require a highly portable AI solution that can run on various platforms with minimal setup

Ready-to-paste prompt

To test the model's capabilities, you can use a prompt like `What are the implications of using a large language model like Kimi K3 in a resource-constrained environment?`

Heads up: Be aware that running a 2.78-trillion-parameter model on a single CPU may require significant computational resources and time, so ensure your system has sufficient RAM (at least 8.24 GB) and a capable CPU to handle the workload

Saves to your device

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

🛠️ Technical setup

Expect 20–40 minutes in a terminal — or let your AI agent drive it.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/kimi-k3-in-c && curl -fsSL https://workflowstacks.com/api/skills/kimi-k3-in-c/claude-skill -o ~/.claude/skills/kimi-k3-in-c/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

How kimi-k3-in-c works

Codeflow

Free to inspect

Kimi-k3-in-c is a large C project (~18k lines across 52 code files, plus 11 test files). It is a full software project: use it through its install path rather than reading it end to end. Last commit this month, Apache-2.0 license, has a test suite.

Size

Large codebase

~18k lines · 52 code files · ~3 h to skim

Setup

Developer setup

A real software project. Use it via its install path; don't expect to read it all.

Runs on

Python

No API keys detected

C 66%Python 25%Shell 5%Makefile 2%CMake 1%

Where to start reading

1
README.md
Start here — what it does and how to install it
2
pyproject.toml
Dependencies and the commands it exposes
3
src/cache/k3_cache.c
Inside src/ — the main logic begins here
4
examples/01-hello.sh
A worked example — copy this to get going

What's in each folder

src/Core code — the actual logic14 files

docs/Documentation180 files

examples/Examples you can copy3 files

tools/Helper scripts23 files

scripts/Helper scripts5 files

benchmarks/Evaluations & benchmarks3 files

cmake/Folder2 files

include/Folder2 files

READMEHas testsDocumentedCI checksExamples includedApache-2.0 licenseUpdated this month

Use with Claude

New

Skip the builder — one click puts this in Claude, Cursor, Antigravity and more.

🛠️ Technical setup

Expect 20–40 minutes in a terminal — or let your AI agent drive it.

Try it instantly — no install

Claude Code

mkdir -p ~/.claude/skills/kimi-k3-in-c && curl -fsSL https://workflowstacks.com/api/skills/kimi-k3-in-c/claude-skill -o ~/.claude/skills/kimi-k3-in-c/SKILL.md

Open in another AI app

Opens the app with this repo with the prompt ready to go — no copy-paste needed.

Connect the whole catalog (MCP)

claude mcp add --transport http workflowstacks https://workflowstacks.com/api/mcp

Adds a WorkflowStacks connector to Claude Code: search and load any skill here by chatting.

Quick Actions

Details

Creator

FareedKhan-dev

Language