Skip to content
localllm
日本語

moonshotai

Kimi-K2.6

Image input

Parameters
1,026.4B
Max context length
262,144
Architecture
deepseek2
License
Conditional
Base model
moonshotai/Kimi-K2.6
License
modified-mit

About Kimi-K2.6

Kimi-K2.6 is a GGUF model based on moonshotai/Kimi-K2.6 on Hugging Face. GGUF files are published by unsloth/Kimi-K2.6-GGUF. It has about 1,026.4B parameters, the deepseek2 architecture, a maximum context length of 262,144 tokens (values written in the GGUF).

There are 4 quantizations, with file sizes from 340 GB (UD-Q2_K_XL) to 2,053 GB (BF16) (actual sizes on Hugging Face).

Running UD-Q2_K_XL (340 GB) with an 8,192-token context needs an estimated 342 GB of memory, and an estimated 344 GB at 32,768 tokens.

The original model is licensed under Modified MIT License (Moonshot AI's Kimi), which allows commercial use with conditions such as scale or purpose. Check the original license for the terms of use. An mmproj file for image input is also available.

File size and estimated memory per quantization

  • Up to 8 GB
  • Up to 16 GB
  • Up to 24 GB
  • Up to 48 GB
  • Up to 128 GB
  • More

File sizes are the actual values on Hugging Face. Memory is estimated for each context length (KV cache f16). The formula is in How memory is calculated.

All memory figures are Estimated (KV cache f16). It should run on a setup with at least this much VRAM (or Mac unified memory). Click a value to see the devices and models that fit that amount (Find by GPU / Mac).

Using image input: This model supports image input. To use images, an mmproj file is loaded as well.

About the "quality guide" (how far a quantization drifts from the original)

For each quantization type, how far the output drifts from the original model, grouped into six levels from values llama.cpp measured on Llama-3-8B.

The value is mean KLD (the difference between the original and quantized model in next-token probabilities); smaller means closer to the original. Quantizations made with an imatrix (importance data) come out slightly smaller. Levels use the value without an imatrix (or with one, for types that only have that).

These values were not measured on this model. Larger models and MoE models tend to drift less with the same type. Distributor-specific mixed types (such as UD-Q4_K_XL) cannot be matched by type name alone, so they show "-".

Quantization Quality guide KLD (no imatrix) KLD (imatrix) Size (Llama-3-8B)
Q8_0 Nearly identical 0.0014 - 7.96 GiB
Q6_K Very small 0.0055 - 6.14 GiB
Q5_K_M Very small 0.0108 - 5.33 GiB
Q5_K_S Small 0.0166 - 5.21 GiB
Q5_1 Small 0.0180 - 5.65 GiB
Q5_0 Small 0.0222 - 5.21 GiB
Q4_K_M Small 0.0313 0.0282 4.58 GiB
Q4_K_S Medium 0.0431 0.0320 4.37 GiB
IQ4_NL Small - 0.0357 4.35 GiB
IQ4_XS Small - 0.0363 4.14 GiB
Q4_1 Medium 0.0717 - 4.78 GiB
Q4_0 Medium 0.0719 - 4.34 GiB
Q3_K_L Medium 0.0872 0.0731 4.03 GiB
Q3_K_M Medium 0.1019 0.0844 3.74 GiB
IQ3_M Medium - 0.1025 3.53 GiB
IQ3_S Medium - 0.1113 3.42 GiB
IQ3_XS Large - 0.1387 3.28 GiB
IQ3_XXS Large - 0.1836 3.05 GiB
Q3_K_S Large 0.2282 0.1998 3.41 GiB
IQ2_M Large - 0.3260 2.74 GiB
Q2_K Very large 0.4451 0.3322 2.96 GiB
Q2_K_S Very large - 0.4034 2.96 GiB
IQ2_S Very large - 0.4392 2.56 GiB
IQ2_XS Very large - 0.5463 2.43 GiB
IQ2_XXS Very large - 0.8120 2.24 GiB
IQ1_M Very large - 1.3931 2.01 GiB
IQ1_S Very large - 2.2546 1.88 GiB

Source: llama.cpp (MIT), measured on Llama-3-8B against FP16, Wikitext-2, revision f364eb6f. Original table · Reading and choosing quantizations

GGUF source

unsloth/Kimi-K2.6-GGUF

Converted and published by a third party About this label

Last checked: Oct 16, 2026

Architecture
deepseek2
Supported before llama.cpp b4409
Parameters
1,026.4B
Max context length
262,144
License
modified-mit Conditional
File size and memory per quantization for Kimi-K2.6 (unsloth/Kimi-K2.6-GGUF)
Quantization File size Measured Quality guide 4,096 8,192 16,384 32,768 65,536 131,072 262,144 max
UD-Q2_K_XL (8 parts) 340 GB - 342 GB 342 GB 343 GB 344 GB 346 GB 351 GB 360 GB
UD-Q4_K_XL (14 parts) 584 GB - 585 GB 585 GB 586 GB 587 GB 589 GB 594 GB 603 GB
UD-Q8_K_XL (14 parts) 595 GB - 596 GB 596 GB 597 GB 598 GB 600 GB 605 GB 614 GB
BF16 (46 parts) 2,053 GB Original precision 2,054 GB 2,055 GB 2,055 GB 2,056 GB 2,059 GB 2,063 GB 2,072 GB

The mmproj file for image input (F16) is 0.95 GB Measured . It is not included in the memory above, so add it if you use images. 

Ollama  ollama run hf.co/unsloth/Kimi-K2.6-GGUF:UD-Q2_K_XL
llama.cpp  llama-server -hf unsloth/Kimi-K2.6-GGUF:UD-Q2_K_XL

Third-party evaluations

Values from other organizations for the original model before quantization. Not this site's own rating. Values are for the original model before quantization. GGUF quantization (especially small quantizations) can lower performance.

Human preference (LMArena)

1,461 1,456–1,465

Kimi K2.6 · 39,573 votes · Values as of Nov 9, 2026

Source: LMArena (CC BY 4.0). Original leaderboard · See in ranking

Capabilities (Epoch AI)

151.5 149.8–153.4

Kimi K2.6 · Values as of Apr 20, 2026

Source: Epoch Capabilities Index (Epoch AI) (CC BY 4.0). Original leaderboard · See in ranking

Hard science (GPQA)

90.8% 87.4%–94.1%

kimi-k2.6 · Values as of May 1, 2026

Source: GPQA Diamond (Epoch AI) (CC BY 4.0). Original leaderboard · See in ranking

License and commercial use

Modified MIT License (Moonshot AI's Kimi) Conditional

MIT License plus one condition. If a commercial product or service has more than 100 million monthly active users or more than US$20 million in monthly revenue, the model name (Kimi K2, Kimi K2.6, Kimi K2.7 Code, etc.) must be prominently displayed in its interface. Otherwise it is like MIT: keep the copyright notice and license text. We checked the original text of Moonshot AI's Kimi K2, K2.6 and K2.7 Code; other authors using the same name may have different terms.

Based on the license on the original model's Hugging Face page. Original license text (checked Nov 1, 2026)

The summary states facts this site read from the original text and is not legal advice. Always check the original license and its latest terms before use. All licenses

How to run

Example commands using UD-Q2_K_XL. To use another quantization, replace the name with one from the table above. How to run local LLMs

Ollama

Run it by pointing directly at the GGUF on Hugging Face.

ollama run hf.co/unsloth/Kimi-K2.6-GGUF:UD-Q2_K_XL

llama.cpp

Pass the repository and quantization name to llama-server (or llama-cli) with -hf.

llama-server -hf unsloth/Kimi-K2.6-GGUF:UD-Q2_K_XL

LM Studio

Paste unsloth/Kimi-K2.6-GGUF into the search box, download it, and choose the quantization from the file list.

Frequently asked questions

How much memory do I need to run Kimi-K2.6 (UD-Q2_K_XL)?
The file size is 340 GB (measured). With a 8,192-token context, the estimated memory including the KV cache is 342 GB. The formula is published at https://localllm.az-arte.com/en/guide/memory-calculation.
Can Kimi-K2.6 be used commercially?
The original model's license is Modified MIT License (Moonshot AI's Kimi) (Commercial use with conditions).MIT License plus one condition. If a commercial product or service has more than 100 million monthly active users or more than US$20 million in monthly revenue, the model name (Kimi K2, Kimi K2.6, Kimi K2.7 Code, etc.) must be prominently displayed in its interface. Otherwise it is like MIT: keep the copyright notice and license text. We checked the original text of Moonshot AI's Kimi K2, K2.6 and K2.7 Code; other authors using the same name may have different terms. Check the original license text before deciding.
Does Kimi-K2.6 run in 8 GB, 12 GB, 16 GB or 24 GB of VRAM?
8 GB: no quantization fits; 12 GB: no quantization fits; 16 GB: no quantization fits; 24 GB: no quantization fits (estimated at 8,192 tokens with an f16 KV cache, files from the most-downloaded source).
What is the command to run Kimi-K2.6 with Ollama?
Run "ollama run hf.co/unsloth/Kimi-K2.6-GGUF:UD-Q2_K_XL" to use the GGUF on Hugging Face directly. To use another quantization, replace the name at the end.