KI

Kimi-K2-Instruct

KimiGGUF by unslothUpdated 2025-12-29
Download on Hugging Face
Parameters1030B
Context128K
Layers61
Architecturedeepseek2
Smallest quant243.6 GB
Quick answer

Kimi-K2-Instruct needs about 622.4 GB of memory at Q4_K_M (a 620.8 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 27 quants, from 243.6 GB (TQ1_0) to 2053.2 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.3 GB.

Kimi-K2-Instruct: 1 trillion‑parameter MoE model for instruction following

Kimi-K2-Instruct is a 1 trillion‑parameter mixture‑of‑experts language model with 32 billion activated parameters. It is trained on 15.5 trillion tokens and optimized for instruction following and agentic tasks. The model is available in multiple GGUF quantizations.

Good at

  • Instruction following
  • Agentic reasoning
  • Coding tasks

Limits

  • Requires large VRAM for high‑bit quants
  • No vision support
  • License is other

Best for

  • Chat‑based agent applications
  • Tool‑use automation
  • Code generation

Who it’s for. Developers and researchers needing a high‑capacity instruction model for building chatbots or autonomous agents. The model is suitable for users with access to high‑memory GPUs or large RAM setups.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
TQ1_0243.6 GB (5 parts) 245.2 GB
Too big Get file →
IQ1_S280.1 GB (6 parts) 281.7 GB
Too big Get file →
IQ1_M304.2 GB (7 parts) 305.8 GB
Too big Get file →
IQ2_XXS328.8 GB (7 parts) 330.4 GB
Too big Get file →
IQ2_M347.1 GB (8 parts) 348.7 GB
Too big Get file →
Q2_K373.2 GB (8 parts) 374.8 GB
Too big Get file →
Q2_K_L373.5 GB (8 parts) 375.0 GB
Too big Get file →
Q2_K_XL381.9 GB (8 parts) 383.4 GB
Too big Get file →
IQ3_XXS416.6 GB (9 parts) 418.2 GB
Too big Get file →
Q3_K_S442.4 GB (10 parts) 444.0 GB
Too big Get file →
Q3_K_XL452.1 GB (10 parts) 453.7 GB
Too big Get file →
Q3_K_M489.4 GB (11 parts) 491.0 GB
Too big Get file →
IQ4_XS546.5 GB (12 parts) 548.1 GB
Too big Get file →
IQ4_NL578.5 GB (12 parts) 580.1 GB
Too big Get file →
Q4_0580.6 GB (13 parts) 582.2 GB
Too big Get file →
Q4_K_S582.7 GB (13 parts) 584.3 GB
Too big Get file →
Q4_K_XL587.1 GB (13 parts) 588.7 GB
Too big Get file →
Q4_K_M620.8 GB (13 parts) 622.4 GB
Too big Get file →
Q4_1642.5 GB (13 parts) 644.1 GB
Too big Get file →
Q5_K_S706.6 GB (15 parts) 708.1 GB
Too big Get file →
Q5_K_M728.3 GB (16 parts) 729.9 GB
Too big Get file →
Q5_K_XL730.6 GB (16 parts) 732.2 GB
Too big Get file →
Q6_K842.6 GB (18 parts) 844.2 GB
Too big Get file →
Q6_K_XL879.1 GB (19 parts) 880.6 GB
Too big Get file →
Q8_01091.1 GB (23 parts) 1092.6 GB
Too big Get file →
Q8_K_XL1190.1 GB (25 parts) 1191.6 GB
Too big Get file →
BF162053.2 GB (45 parts) 2054.7 GB
Too big Get file →

For 8 GB VRAM, use Q4_0 or Q4_K_S (file size ~580 GB, memory ~582 GB). For 16 GB VRAM, Q4_K_M (~620 GB) is recommended. If you have 32 GB VRAM, consider Q5_K_S (~706 GB) for better quality. Memory estimates are approximate.

What GPU do you need for Kimi-K2-Instruct?

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-K2-Instruct quant should you choose?

Q4_K_M is the usual starting point for Kimi-K2-Instruct: it balances file size and quality well. Here it is a 620.8 GB file.

The Q8_0 file is 1091.1 GB, about 1.8 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.

If memory is tight, the smallest file is TQ1_0 at 243.6 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architecturedeepseek2
Parameters1030B (approximate, mixture of experts)
Layers61
Trained context131,072 tokens
KV cache (fp16)130 KB per token

Source

Quantized byunsloth
FamilyKimi
Licenseother
Downloads (HF)15,648
Last updated2025-12-29

Frequently asked questions about Kimi-K2-Instruct

How much VRAM does Kimi-K2-Instruct need?

About 622.4 GB at Q4_K_M with an 8K context window. That is the 620.8 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-K2-Instruct on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 245.2 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Kimi-K2-Instruct GGUF?

It depends on the quant: TQ1_0 is 243.6 GB, IQ1_S is 280.1 GB, IQ1_M is 304.2 GB, IQ2_XXS is 328.8 GB, IQ2_M is 347.1 GB, Q2_K is 373.2 GB and 21 more.

What context length does Kimi-K2-Instruct support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 1.1 GB at 8K and 4.3 GB at 32K.

Where can I download Kimi-K2-Instruct GGUF?

The files are on Hugging Face in the unsloth/Kimi-K2-Instruct-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-K2-Instruct 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-K2-Instruct on 8 GB VRAM?

Yes, but only with low‑bit quants such as Q4_0 or Q4_K_S, which require about 580 GB of storage and ~582 GB of memory at 8 k context.

Does the model support vision inputs?

No, the model does not have vision capabilities.

What is the recommended temperature setting?

A temperature of 0.6 is suggested to reduce repetition and incoherence.

Other GGUFs of Kimi-K2-Instruct

Similar models

Browse all →

More from unsloth

All by unsloth →

More Kimi models

All Kimi →

Other large models

All large models →