KI

Kimi K3

KimiGGUF by AtomicChatUpdated 2026-10-28
Download on Hugging Face
Parameters2780B
Context1024K
Layers93
Architecturekimi-k3
Smallest quant589.5 GB
Quick answer

Kimi K3 needs about 938.1 GB of memory at IQ2_XXS (a 769.1 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 5 quants, from 589.5 GB (IQ1_S) to 2955.0 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 674.0 GB.

Kimi K3 – a 2.8T parameter multimodal model with 1M token context

Kimi K3 is a 2.8T parameter mixture‑of‑experts model built on the Kimi‑Delta Attention architecture. It supports text, images, and video and offers a 1‑million‑token context window. The model is released under the Kimi K3 License for research and deployment.

Good at

  • Long‑horizon coding
  • Native multimodal reasoning
  • Large context handling

Limits

  • Requires >500 GB of fast memory
  • No vision support on older llama.cpp builds
  • License is other, not standard

Best for

  • Research on multimodal LLMs
  • Long‑form content generation
  • Complex coding assistance

Who it’s for. Researchers and developers with large‑scale GPU or RAM resources who need a multimodal model with a massive context window. The model is suitable for advanced experimentation and prototype development in environments that can accommodate its memory footprint.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S589.5 GB (14 parts) 758.6 GB
Too big Get file →
IQ1_M689.9 GB (16 parts) 858.9 GB
Too big Get file →
IQ2_XXS769.1 GB (18 parts) 938.1 GB
Too big Get file →
Q3_K_S1197.0 GB (28 parts) 1366.0 GB
Too big Get file →
Q8_02955.0 GB (71 parts) 3124.0 GB
Too big Get file →

For 8 GB VRAM, the smallest quant IQ1_S (~589.54 GB) is still too large; a local run is not feasible. If you have a server with at least 600 GB of combined RAM and VRAM, IQ1_S is the most memory‑efficient option. For larger memory budgets, IQ1_M or IQ2_XXS provide better precision while remaining within the 700–800 GB range.

What GPU do you need for Kimi K3?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: no quant fits at an 8K context.
  • 32 GB VRAM: no quant fits at an 8K context.
  • 48 GB VRAM: no quant fits at an 8K context.

Which Kimi K3 quant should you choose?

IQ2_XXS (769.1 GB) is the middle option in this repository. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

If memory is tight, the smallest file is IQ1_S at 589.5 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturekimi-k3
Parameters2780B (approximate, mixture of experts)
Layers93
Trained context1,048,576 tokens
KV cache (fp16)20088 KB per token

Source

Quantized byAtomicChat
FamilyKimi
Licenseother
Downloads (HF)3,425
Last updated2026-10-28

Frequently asked questions about Kimi K3

How much VRAM does Kimi K3 need?

About 938.1 GB at IQ2_XXS with an 8K context window. That is the 769.1 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Kimi K3 on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 758.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Kimi K3 GGUF?

It depends on the quant: IQ1_S is 589.5 GB, IQ1_M is 689.9 GB, IQ2_XXS is 769.1 GB, Q3_K_S is 1197.0 GB, Q8_0 is 2955.0 GB.

What context length does Kimi K3 support?

The GGUF header lists a trained context length of 1,048,576 tokens (1024K). The KV cache grows with the context you use: about 168.5 GB at 8K and 674.0 GB at 32K.

Where can I download Kimi K3 GGUF?

The files are on Hugging Face in the AtomicChat/Kimi-K3-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Kimi K3 use?

Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Can I run Kimi K3 on a laptop with 8 GB VRAM?

No. The smallest quant weighs about 589.54 GB, far exceeding typical laptop memory limits.

Does Kimi K3 support vision?

Yes. It includes a native vision encoder, MoonViT‑V2, and supports image and video input.

What license applies to Kimi K3?

The model is released under the Kimi K3 License, which is not a standard open‑source license.

Other GGUFs of Kimi-K3

Similar models

Browse all →

More from AtomicChat

All by AtomicChat →

More Kimi models

All Kimi →

Other large models

All large models →