LLMs ·

Kimi K3 (2.8T)

Parameters
2,780B (104.0B active)
Architecture
Mixture of experts
Max context
1,024K tokens
License
kimi-k3
Released
2026-07

How to run it

Ollama

Ollama offers this only as a cloud model (it runs on Ollama’s servers, not your machine). To run it locally, use the GGUF files below.

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/unsloth/Kimi-K3-GGUF (opens in a new tab)

ollama run hf.co/unsloth/Kimi-K3-GGUF

Original weights · Hugging Face

Full-precision files from Moonshot AI, for fine-tuning, vLLM, Transformers.

huggingface.co/moonshotai/Kimi-K3 (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Moonshot AI. The weights themselves download from Hugging Face.

github.com/MoonshotAI/Kimi-K3 (opens in a new tab)

git clone https://github.com/MoonshotAI/Kimi-K3

Links checked 2 days ago against Hugging Face, Ollama and GitHub.

Apps that run Kimi K3 (2.8T)

Our pick for each kind of computer, assuming the model fits in its memory. Pick your own GPU or device in the Model Finder for an exact answer.

Mac (Apple silicon)

LM Studio

Point-and-click: search, download and chat in one window.

Also good:

Windows PC, NVIDIA GPU

LM Studio

Point-and-click: search, download and chat in one window.

Also good:

Windows PC, AMD Radeon

Lemonade

AMD’s own app: llama.cpp tuned for Radeon on Vulkan and ROCm.

Also good:

Linux, NVIDIA GPU

LM Studio

Point-and-click: search, download and chat in one window.

Also good:

Linux server, 80 GB+ GPUs

LM Studio

Point-and-click: search, download and chat in one window.

Also good:

App Type Opens it as Systems Engine Start
LM Studio Desktop app GGUF macOSWindowsLinux llama.cpp, MLX Open in LM Studio
Ollama CLI + API GGUF from Hugging Face macOSWindowsLinux llama.cpp, MLX 
ollama run hf.co/unsloth/Kimi-K3-GGUF:Q4_K_M
llama.cpp CLI + API GGUF macOSWindowsLinux llama.cpp 
llama-server -hf unsloth/Kimi-K3-GGUF:Q4_K_M
Jan Desktop app GGUF macOSWindowsLinux llama.cpp, MLX Open in Jan
vLLM Server Original weights WSLWindowsLinux vLLM 
vllm serve moonshotai/Kimi-K3
SGLang Server Original weights WSLWindowsLinux SGLang 
python3 -m sglang.launch_server --model-path moonshotai/Kimi-K3
Docker Model Runner CLI + API GGUF macOSWindowsLinux llama.cpp, vLLM 
docker model run hf.co/unsloth/Kimi-K3-GGUF:Q4_K_M
Lemonade Desktop app GGUF macOSWindowsLinux llama.cpp, ONNX Runtime GenAI 
lemonade pull unsloth/Kimi-K3-GGUF:Q4_K_M
Unsloth Desktop Desktop app GGUF macOSWindowsLinux llama.cpp, MLX, stable-diffusion.cpp Open in Unsloth Desktop
Atomic Chat Desktop app GGUF macOSWindowsLinux llama.cpp, MLX-VLM Open in Atomic Chat
KoboldCpp Desktop app GGUF macOSWindowsLinux llama.cpp Get KoboldCpp
Text Generation WebUI Desktop app GGUF macOSWindowsLinux llama.cpp, Transformers, ExLlamaV3 Get Text Generation WebUI
llamafile CLI + API GGUF macOSWindowsLinux llama.cpp Get llamafile
Msty Studio Desktop app GGUF macOSWindowsLinux Ollama, llama.cpp, MLX Open in Msty Studio
Llama (macOS) Desktop app GGUF macOS llama.cpp Open in Llama (macOS)
RecurseChat Desktop app GGUF macOS llama.cpp Open in RecurseChat
node-llama-cpp Library GGUF macOSWindowsLinux llama.cpp Get node-llama-cpp
PocketPal AI Mobile app GGUF iOSAndroid llama.cpp App Store Google Play
ChatterUI Mobile app GGUF Android llama.cpp Download the APK
LLM Farm Mobile app GGUF iOS llama.cpp Not on the App Store or TestFlight for now (checked 2026-11-09).

Which apps can open a model depends on the files published for it: GGUF for llama.cpp-based apps, MLX for Apple’s engine, the original weights for vLLM and SGLang. Compare every app.

Memory needed

Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps. Try another context length.

KV cache: hybrid linear attention: about 1 in 4 layers keeps a latent (MLA) cache; active parameters as published.

Quantization Quality Weights Total at 8K context Total at 32K context Download
Q2_K Low 1,112 GB 1,113 GB 1,113 GB GGUF (UD-Q2_K_XL) · folder (opens in a new tab)
Q3_K_M Low 1,362 GB 1,363 GB 1,364 GB GGUF files (opens in a new tab)
Q4_K_M Good 1,696 GB 1,697 GB 1,697 GB GGUF (UD-Q4_K_XL) · folder (opens in a new tab)
Q5_K_M High 1,974 GB 1,975 GB 1,975 GB GGUF files (opens in a new tab)
Q6_K High 2,280 GB 2,280 GB 2,281 GB GGUF files (opens in a new tab)
Q8_0 Very high 2,947 GB 2,948 GB 2,948 GB GGUF (UD-Q8_K_XL) · folder (opens in a new tab)
F16 Full 5,560 GB 5,561 GB 5,561 GB Original weights (opens in a new tab)

Best hardware for Kimi K3 (2.8T)

None. No single GPU or device in our catalog runs this model on its own; it needs several GPUs working together.

Check it on your own hardware

Run it on a cluster

Devices that can’t run Kimi K3 (2.8T) alone can run it together. These example clusters do:

Try it with your own devices

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.

Was this page helpful?