[whatmodel.sh](../../README.md) / Kimi K2.6
# Kimi K2.6
Moonshot AI · released 2026-04 · 1000B params (32B active) · modified-mit
**score 71.5** ±0.5 · #64 of 732 · from 56 benchmark results across 17 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Kimi K2.6 among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█▄▅▅●▄▂▁ #58 of 313 20% 13
coding ▁▁▁▃▅▇█▇▅●▃▁ #54 of 469 18% 7
chat & writing ▁▁▁▃▆▄▆█▆▇●▃ #43 of 402* 2.4% 1
math ▁▁▃▅█▆▄█●▂▁▁ #45 of 431 30% 11
vision & documents ▁▁▂▂▁▃▄▂▆▇●▄ #39 of 153 3.5% 4
web search ▁▁▁●▄█▃▄·▁·▁ #60 of 84 2.9% 3
agent tasks · #58 of 313 · 20% of its score
Beats [Muse Spark 1.1](muse-spark-1-1.md), [Muse Spark 1.2](muse-spark-1-2.md), [Qwen3.7 Max](qwen3-7-max.md) here, though they rank higher overall. Trails [Beam 1.0 501B A23B](beam-1-0-501b-a23b.md), [Claude Sonnet 4.6](claude-sonnet-4-6.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
SWE-bench Verified 76.7% ▁··▁·▁▂▁▄●▄▁ beats 22/29 hard · avg 19% 7.0% sharp · several runs
Terminal-Bench 2.1 67.2% █··▄█·▄····● beats 6/6 very hard · avg 13% 2.0% sharp · at its level
SWE-bench Multilingual 77.1% ▄··▄▄···▄▄·● beats 6/6 hard · avg 30% 1.9% typical
Agents' Last Exam overall full-pa… 9.2% ▄···█●▄▄█▄·█ beats 3/10 very hard · avg 1.9% 1.7% typical
Humanity's Last Exam with tools 54.0% ▄····▄·█·▄·● beats 5/5 hard · avg 15% 1.2% at its level
Thinkingbox-bench pass@1 (507 sta… 37.7% ▅··▃·▃·█●▅██ beats 7/15 very hard · avg 5.1% 1.0% sharp · at its level
BrowseComp 61.3% ▄·▄▄···▄··█● beats 6/6 hard · avg 18% 1.0% at its level
τ³-bench 72.4% ▄····▄·█·●·█ beats 4/6 medium · avg 42% 0.9% at its level
Legal Research Bench (agentic, al… 15.5% ▄●··▄▄▄██·▄█ beats 2/12 very hard · avg 6.3% 0.7% at its level
EEGAgentBench overall (tool-based… 51.3% ▂·▁▁··▇▇●▂·▁ beats 21/25 hard · avg 28% 0.7% blunt · at its level
KlinikeBench strict pass@1 (333 i… 23.4% ▁▁·▁▁▁▁▅▃█▅● beats 22/27 very hard · avg 2.4% 0.6% typical
TerminalWorld-Verified (200 tasks… 57.5% ▄··▄·▄·●···▄ beats 4/5 hard · avg 28% 0.5% blunt · at its level
τ³-bench telecom (tau2-bench ≥ v1… 97.8% ▄·▄···▄·▄·●▄ beats 4/6 easy · avg 79% 0.5% maxed out
coding · #54 of 469 · 18% of its score
Beats [GPT-5.4](gpt-5-4.md), [Gemini 3.1 Pro (Preview)](gemini-3-1-pro.md), [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md) here, though they rank higher overall. Trails [Qwen3.7 Plus](qwen3-7-plus.md), [Claude Sonnet 4.6](claude-sonnet-4-6.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
SWE-bench Verified 76.7% ▁··▁·▁▂▁▄●▄▁ beats 22/29 hard · avg 19% 7.0% sharp · several runs
Arena WebDev (Code Arena) 1509.2 ▁▁▃▂▅█▃●▅▅▁▂ beats 74/129 avg 981 4.2% sharp
SWE-bench Multilingual 77.1% ▄··▄▄···▄▄·● beats 6/6 hard · avg 30% 1.9% typical
SciCode subproblems 52.0% ▃····▃··█·▃● beats 6/6 hard · avg 19% 1.2% at its level
SciCode (level unspecified) 53.5% ▄·····▄█▄▄·● beats 6/6 hard · avg 16% 1.1% at its level
LiveCodeBench v6 (window unspecif… 90.2% ██···█··█●██ beats 4/6 medium · avg 46% 1.1% typical
LiveCodeBench v6 (Feb–May 2025) 86.5% ▄··█····▄▄·● beats 6/6 medium · avg 47% 0.9% typical
chat & writing · #43 of 402 · 2.4% of its score
Beats [GPT-6 Sol](gpt-6-sol.md), [MiMo-V2.6-Flash (309B-A15B)](mimo-v2-6-flash.md), [Claude Haiku 5.5](claude-haiku-5-5.md) here, though they rank higher overall. Trails [MiMo-V2.5-Pro (1T-A42B)](mimo-v2-5-pro.md), [GLM-5.1](glm-5-1.md), [Grok 4.20 (0309 reasoning)](grok-4-20.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1461.0 ▁▁▁▂▃▃▃▆▆▇●▃ beats 335/385 avg 1260 2.4% blunt
math · #45 of 431 · 30% of its score
Beats [Grok 4.7](grok-4-7.md), [GLM-5.3-Flash](glm-5-3-flash.md), [Gemini 3.6 Flash](gemini-3-6-flash.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
FrontierMath tiers 1–3 v2 57.2% ▂··▁▁▃▄▄●▆▄▃ beats 42/86 very hard · avg 1.2% 11% sharp · at its level
FrontierMath tier 4 v2 25.6% ▁▂▁▂●▅▂▃▂▁▁▁ beats 27/67 very hard · avg 0.1% 7.1% sharp
MathArena ArXivMath February 2026… 43.0% ▁▁▁▁█▇●▃▁▃·▃ beats 15/24 very hard · avg 9.4% 2.8% at its level
MathArena ArXivMath March 2026 (3… 55.8% █▃▅·●▃▃·▅·▃▃ beats 7/14 hard · avg 16% 2.2% at its level
MathArena ArXivMath January 2026 … 71.7% ▂··▄▂▄█▆▆▄●█ beats 19/23 hard · avg 23% 1.9% at its level
IMO-AnswerBench 91.1% ▄····▄·▄▄·●▄ beats 5/6 medium · avg 41% 0.9% typical
MathArena BrokenArXiv March 2026 … 15.6% █·▆●▂·▂▂···▂ beats 9/14 very hard · avg 1.0% 0.8% sharp · noisy
MathArena BrokenArXiv February 20… 11.7% ▃▂▂▃●▃▂·▃··▂ beats 7/16 very hard · avg 0.9% 0.8% sharp · noisy
AIME 2026 95.4% ▄·▄·▄▄··●··▄ beats 4/6 medium · avg 57% 0.7% sharp · maxed out · noisy
OTIS mock AIME 2024–2025 96.1% ▃▃▄▃▃▃▇▇█●▂▅ beats 167/197 very hard · avg 7.2% 0.6% sharp · maxed out · noisy
MathConstraint full release, no t… 35.9% ▄··▄▄·●▄█▄·▄ beats 3/8 very hard · avg 11% 0.5% at its level
vision & documents · #39 of 153 · 3.5% of its score
Beats [MiMo-V2.6-Pro](mimo-v2-6-pro.md), [MiMo-V2.6-Flash (309B-A15B)](mimo-v2-6-flash.md), [DeepSeek V4.1 Flash](deepseek-v4-1-flash.md) here, though they rank higher overall. Trails [Qwen3.6 Flash](qwen3-6-flash.md), [Qwen3.6 Plus](qwen3-6-plus.md), [Qwen3.5 397B-A17B](qwen3-5-397b-a17b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena vision (style controlled) 1264.6 ▁▁▂▂▂▂▃▄▅▅●▇ beats 119/152 avg 1061 1.6% blunt
CharXiv (reasoning questions) (ma… 80.4% ▂▅▃▅▅▆▆▅▆●▆▂ beats 79/101 medium · avg 38% 1.1% small run
Arena document (style controlled) 1449.8 ▁▁▃▃▄▃●█▅▇▄▅ beats 11/37 avg 1331 0.8% blunt
PureDocBench document parsing (Av… 70.1% ▂·▁·▂▂▅●·▁▂▂ beats 15/22 medium · avg 62% 0.0% blunt · at its level
web search · #60 of 84 · 2.9% of its score
Trails [GPT-4o mini (2024-07-18)](gpt-4o-mini-2024-07-18.md), [Mistral Small 3.1 24B (2503)](mistral-small-3-1.md), [Llama 4 Scout (109B-A17B)](llama-4-scout.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Humanity's Last Exam with tools 54.0% ▄····▄·█·▄·● beats 5/5 hard · avg 15% 1.2% at its level
BrowseComp 61.3% ▄·▄▄···▄··█● beats 6/6 hard · avg 18% 1.0% at its level
Legal Research Bench (agentic, al… 15.5% ▄●··▄▄▄██·▄█ beats 2/12 very hard · avg 6.3% 0.7% at its level
other tests (19)
test result vs the others in its run difficulty counts why
GPQA Diamond 90.8% ▁▁▁▁▂▄▆▄▅▇●▅ beats 179/211 hard · avg 44% 5.4% sharp · several runs
Humanity's Last Exam 34.8% ▄··█▄··▄·●·▄ beats 5/6 very hard · avg 5.7% 4.6% at its level · several runs
Furniture Assembly (Epoch) 21.7% ●▆▆▆▆▂▆▂▂▂▂▅ beats 2/32 very hard · avg 1.2% 3.4% sharp · at its level
EBR-bench (Epoch) 2.4% ●▂·▆█▂█▆▄▆·▆ beats 1/25 very hard · avg 0.3% 3.1% sharp
GPQA (subset unspecified) 91.0% ▄█▄▄·▄·····● beats 6/6 hard · avg 47% 3.0% sharp
Mystery Game Puzzles (Epoch) 18.0% █··▂▃●▄█▂▄▂▁ beats 30/77 very hard · avg 9.4% 2.6% sharp · noisy
MMLU-Pro 88.1% ▄·····▄▄▄·▄● beats 5/6 medium · avg 53% 1.9% typical
Chess Puzzles (Epoch) 26.0% █▁▁▁▁▂▃●▃▂▁▁ beats 96/137 very hard · avg 5.8% 1.8% sharp · noisy
Global-MMLU-Lite 90.6% ▃······▅··●▃ beats 4/6 medium · avg 61% 1.6% typical
EQ-Bench 3 (emotional intelligenc… 1319.2 ▁·▂▁▄▅▅▃█●▁▁ beats 55/62 avg 1034 1.4% typical
AgentWorldBench (open-ended, over… 53.4% ▄▄··▄▄·●█▄██ beats 4/11 hard · avg 26% 1.2% blunt · at its level
IFBench (aggregation unspecified) 76.0% ▃······●▃▃·▃ beats 3/6 medium · avg 45% 1.0% typical
MMLU-ProX 85.0% ▄·····▄·▄●█▄ beats 3/6 medium · avg 58% 1.0% typical
SimpleQA Verified 34.9% ▁▁▃▂▃▃●▆█▄▂▅ beats 30/81 very hard · avg 8.0% 0.9% at its level
IFBench prompt-level (loose) 73.7% ●▄·▄··▄▄···█ beats 0/6 medium · avg 47% 0.9% typical
Legal Research Bench (single-shot… 2.9% ▃▃●▃·▅···█▃▃ beats 2/12 very hard · avg 0.8% 0.9% too hard for it
MultiChallenge 63.1% ▁··········● beats 2/6 hard · avg 28% 0.9% at its level
CritPt 9.1% ▄····▄█···●▄ beats 4/6 very hard · avg 1.1% 0.5% too hard for it · noisy
Vectara hallucination leaderboard… 89.2% ▁▂▁▂▅●▂▃▄▃▁▁ beats 34/98 very easy · avg 90% 0.0% blunt
source results share of the score
Epoch AI 10 37% ▇▇▇▇
NVIDIA 17 24% ▇▇▇
A.X K2 Technical Report 7 11% ▇
LMArena leaderboard dataset 4 9% ▇
MathArena ArXivMath / BrokenArXiv / ArX… 5 9% ▇
Agents' Last Exam 1 2% ▇
Legal Research Bench 2 2% ▇
EQ-Bench 3 canonical leaderboard results 1 1% ▇
Qwen/Qwen-AgentWorld-35B-A3B model card 1 1% ▇
Moonshot AI vendor-reported benchmark s… 1 1% ▇
One Success Isn't Reliability: Thinking… 1 1% ▇
EEGAgentBench: Benchmarking LLM Agents … 1 1% ▇
KlinikeBench: Evaluating Language Model… 1 1% ▇
MathConstraint: Automated Generation of… 1 1% ▇
TerminalWorld: Benchmarking Agents on R… 1 0% ▇
How Far Is Document Parsing from Solved… 1 0% ▇
Vectara hallucination leaderboard 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 71.5
Licences: Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · NVIDIA ([CC BY 4.0](../../licenses/report-2606.15007v1.txt)) · A.X K2 Technical Report ([CC BY 4.0](http://creativecommons.org/licenses/by/4.0/)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · MathArena ArXivMath / BrokenArXiv / ArXivLean model-output datasets ([CC BY-SA 4.0](../../licenses/matharena-sharealike-NOTICE.txt)) · Agents' Last Exam ([CC BY 4.0](../../licenses/paper-2606.05405v2-CC-BY-4.0.txt)) · Legal Research Bench ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · EQ-Bench 3 canonical leaderboard results ([MIT](../../licenses/eqbench3-MIT.txt)) · Qwen/Qwen-AgentWorld-35B-A3B model card ([Apache-2.0](https://huggingface.co/Qwen/Qwen-AgentWorld-35B-A3B/blob/60d2b0434a53d2e62a7c00a489586815d94ebffb/README.md)) · One Success Isn't Reliability: Thinkingbox ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · EEGAgentBench: Benchmarking LLM Agents on Short- and Long-Horizon EEG Analysis ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · KlinikeBench: Evaluating Language Models Beyond Diagnostic Accuracy ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · MathConstraint: Automated Generation of Verified Combinatorial Reasoning Instances for LLMs ([CC BY 4.0](../../licenses/paper-2605.08498v1-CC-BY-4.0.txt)) · TerminalWorld: Benchmarking Agents on Real-World Terminal Tasks ([CC BY 4.0](../../licenses/paper-2605.22535v1-CC-BY-4.0.txt)) · How Far Is Document Parsing from Solved? PureDocBench: A Source-Traceable Benchmark across Clean ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Vectara hallucination leaderboard ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
## How to run it
### Through an API
api (Moonshot AI) · $1.71/M blended · $0.95 in / $4 out per 1M tokens · [platform.moonshot.ai](https://platform.moonshot.ai/docs/api/chat)
[price source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/providers/moonshotai/models/kimi-k2.6.toml): Standard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ. · [context source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/models/moonshotai/kimi-k2.6.toml)
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run Kimi K2.6 (ud-q4_k_xl, needs ~600.5 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/Kimi-K2.6-GGUF/tree/main))
```sh
llama cli -hf unsloth/Kimi-K2.6-GGUF:UD-Q4_K_XL
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf unsloth/Kimi-K2.6-GGUF:UD-Q4_K_XL
```
Other versions: ud-q2_k_xl (~300.5 GB) `llama cli -hf unsloth/Kimi-K2.6-GGUF:UD-Q2_K_XL`
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run Kimi K2.6 (ud-q4_k_xl, needs ~600.5 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/Kimi-K2.6-GGUF/tree/main))
```sh
lms get https://huggingface.co/unsloth/Kimi-K2.6-GGUF@UD-Q4_K_XL -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: ud-q2_k_xl (~300.5 GB) `lms get https://huggingface.co/unsloth/Kimi-K2.6-GGUF@UD-Q2_K_XL -y` · 4bit (~657.6 GB, Mac only) `lms get https://huggingface.co/aidiffuser/Kimi-K2.6-MLX-4bit -y`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run Kimi K2.6 (4bit, needs ~657.6 GB, Mac only, from [huggingface.co/aidiffuser](https://huggingface.co/aidiffuser/Kimi-K2.6-MLX-4bit/tree/f3526deda33ce0363fc929dad500638c2ae620ac))
```sh
mlx_lm.chat --model aidiffuser/Kimi-K2.6-MLX-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model aidiffuser/Kimi-K2.6-MLX-4bit
```
Model repository and documentation: [huggingface.co/moonshotai](https://huggingface.co/moonshotai/Kimi-K2.6)
[interactive version](https://whatmodel.sh/model/kimi-k2-6) · [JSON](https://whatmodel.sh/api?format=json&all=1)