[whatmodel.sh](../../README.md) / Kimi K3
# Kimi K3
Moonshot AI · released 2026-07 · 2800B params (104B active) · kimi-k3-license
**score 75.4** ±0.5 · #30 of 732 · from 31 benchmark results across 13 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks Kimi K3 among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█▄▅▅▄●▂▁ #31 of 313 17% 6
coding ▁▁▁▃▅▇█▇▅▄●▁ #18 of 469 12% 4
chat & writing ▁▁▁▃▆▄▆█▆▇▇● #15 of 402* 3.2% 1
math ▁▁▃▅█▆▄█▄●▁▁ #26 of 431 48% 10
vision & documents ▁▁▂▂▁▃▄▂▆▇●▄ #24 of 153* 1.2% 1
agent tasks · #31 of 313 · 17% of its score
Beats [Hy4 preview (770B-A49B)](hy4-preview.md), [GPT-5.6 Terra](gpt-5-6-terra.md), [GPT-5.4](gpt-5-4.md) here, though they rank higher overall. Trails [Grok 4.7](grok-4-7.md), [GLM-5.3](glm-5-3.md), [GPT-6 Luna](gpt-6-luna.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena Agent (October 2026 aggrega… 0.0108 IPS ▁▁▁▃▃▂●▁▂▂▁▁ beats 29/59 avg -0.379 IPS 8.1% sharp
GameLogicBench solve rate (72 God… 36.1% █▄·█·▄·●█▄·▄ beats 6/11 very hard · avg 5.1% 2.3% at its level
τ³-bench banking_knowledge (v1.0.… 37.1% ▃▃▂▃·▆▃▂●▅▅▂ beats 14/25 very hard · avg 4.3% 2.2% at its level
E2E-SWE (build a working codebase… 42.3% ▃··▅··▃●▃▅▃▅ beats 4/12 very hard · avg 3.7% 1.9% sharp · at its level
Thinkingbox-bench pass@1 (507 sta… 57.4% ▅··▃·▃·█▃▅●█ beats 10/15 very hard · avg 5.1% 1.4% sharp · at its level
Video2World V2WScore (coding agen… 33.2% █··▄█·●····▄ beats 5/6 very hard · avg 7.3% 1.3% at its level
coding · #18 of 469 · 12% of its score
Beats [GPT-5.6 Sol](gpt-5-6-sol.md), [Claude Fable 5](claude-fable-5.md), [GPT-5.5](gpt-5-5.md) here, though they rank higher overall. Trails [Qwen3.8 Max (0902)](qwen3-8-max-0902.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena WebDev (Code Arena) 1656.4 ▁▁▃▂▅█▃▆▅●▁▂ beats 114/129 avg 981 5.6% sharp
GameLogicBench solve rate (72 God… 36.1% █▄·█·▄·●█▄·▄ beats 6/11 very hard · avg 5.1% 2.3% at its level
Luau Leetcode fully solved (Torpe… 77.9% ▁▁▁▁·▁▂▃▅▃●▇ beats 28/49 hard · avg 17% 2.1% sharp
E2E-SWE (build a working codebase… 42.3% ▃··▅··▃●▃▅▃▅ beats 4/12 very hard · avg 3.7% 1.9% sharp · at its level
chat & writing · #15 of 402 · 3.2% of its score
Beats [GPT-6 Astra](gpt-6-astra.md), [GPT-6.1 Sol](gpt-6-1-sol.md), [Claude Sonnet 5.5](claude-sonnet-5-5.md) here, though they rank higher overall. Trails [Muse Spark 1.2](muse-spark-1-2.md), [Muse Spark 1.1](muse-spark-1-1.md), [Claude Opus 4.6](claude-opus-4-6.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1488.2 ▁▁▁▂▃▃▃▆▆▇█● beats 370/385 avg 1260 3.2% blunt
math · #26 of 431 · 48% of its score
Beats [Gemini 3.8 Flash](gemini-3-8-flash.md), [DeepSeek V4.1 Flash](deepseek-v4-1-flash.md) here, though they rank higher overall. Trails [GPT-5.6 Luna](gpt-5-6-luna.md), [GPT-6 Luna](gpt-6-luna.md), [Claude Haiku 5.5](claude-haiku-5-5.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
FrontierMath tier 4 v2 39.0% ▁▂▁▂█●▂▃▂▁▁▁ beats 43/67 very hard · avg 0.1% 14% sharp · at its level
FrontierMath tiers 1–3 v2 72.2% ▂··▁▁▃▄▄█●▄▃ beats 63/86 very hard · avg 1.2% 13% sharp · at its level
MathArena ArXivMath June 2026 (48… 71.5% ▂···▄·██●▆▄▂ beats 12/19 very hard · avg 5.1% 6.5% sharp · at its level
MathArena ArXivMath April 2026 (4… 54.2% ▁·····▅●█▄·▁ beats 10/20 very hard · avg 7.6% 6.3% sharp · at its level
MathArena BrokenArXiv April 2026 … 59.4% ▂▂▆█▃▃●▂▂··▂ beats 15/18 very hard · avg 0.8% 2.5% sharp · at its level · noisy
MathArena ArXivMath May 2026 (40 … 61.7% ▁··▁▄█●▁·▃·▁ beats 11/17 very hard · avg 11% 2.1% at its level
MathArena BrokenArXiv June 2026 (… 51.9% ▄··█▂▂▂●▄··▂ beats 11/14 very hard · avg 1.3% 1.7% sharp · at its level · noisy
MathArena ArXivMath August 2026, … 38.6% ●▄·▄·······█ beats 0/5 very hard · avg 5.9% 1.2% sharp · at its level
MathArena BrokenArXiv August 2026… 61.9% ▄▄·····●·▄·█ beats 2/5 very hard · avg 6.9% 0.6% at its level · noisy
OTIS mock AIME 2024–2025 97.2% ▃▃▄▃▃▃▇▇█▃●▅ beats 169/197 very hard · avg 7.2% 0.3% sharp · maxed out · noisy
vision & documents · #24 of 153 · 1.2% of its score
Beats [Claude Sonnet 5.5](claude-sonnet-5-5.md), [GPT-6 Sol](gpt-6-sol.md), [MiMo-V2.6-Pro](mimo-v2-6-pro.md) here, though they rank higher overall. Trails [Gemini 3.6 Flash](gemini-3-6-flash.md), [Gemini 3.5 Flash](gemini-3-5-flash.md), [Muse Spark 1.2](muse-spark-1-2.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
CharXiv (reasoning questions) (ma… 84.8% ▂▅▃▅▅▆▆▅▆█●▂ beats 94/101 medium · avg 38% 1.2% small run
other tests (11)
test result vs the others in its run difficulty counts why
Furniture Assembly (Epoch) 34.2% █▆●▆▆▂▆▂▂▂▂▅ beats 12/32 very hard · avg 1.2% 5.5% sharp · at its level
Mystery Game Puzzles (Epoch) 26.0% █··▂▃▇●█▂▄▂▁ beats 44/77 very hard · avg 9.4% 5.5% sharp · at its level · noisy
Chess Puzzles (Epoch) 39.0% █▁▁▁▁▂▃▃●▂▁▁ beats 117/137 very hard · avg 5.8% 3.1% sharp · at its level · noisy
GPQA Diamond 93.1% ▁▁▁▁▂▄▆▄▅▇█● beats 194/211 hard · avg 39% 1.9% sharp
SimpleQA Verified 50.6% ▁▁▃▂▃▃▆▆●▄▂▅ beats 56/81 very hard · avg 8.0% 1.2% at its level
Icelandic LLM Leaderboard average… 40.7% ▁▁▃▂▂▄▁●▄▄▄█ beats 14/38 very hard · avg 14% 1.2% at its level
AutomationBench public set (Zapie… 46.7% ▄······▄●··▄ beats 3/4 very hard · avg 11% 1.0% at its level
Video-MME v2 (with subtitles) 65.4% █·········█● beats 2/2 hard · avg 17% 0.8% at its level · small run
SimpleVQA with tools (search) 71.9% █·········●█ beats 1/2 hard · avg 27% 0.8% at its level · small run
RobloxQA v2.0 multiple choice (To… 96.5% ▁▁▃▁█▄▃▃●▂▂▂ beats 34/49 easy · avg 77% 0.7% maxed out
IntPhys 2 60.3% █···●······█ beats 1/2 hard · avg 27% 0.6% at its level · small run
source results share of the score
Epoch AI 8 44% ▇▇▇▇▇
MathArena ArXivMath / BrokenArXiv / ArX… 7 21% ▇▇▇
LMArena leaderboard dataset 3 17% ▇▇
Roblox LLM Leaderboard 2 3% ▇
GameLogicBench: Evaluating Coding Agent… 1 2% ▇
τ²/τ³-bench leaderboard: Sierra's own e… 1 2% ▇
Meta vendor-reported benchmark scores f… 3 2% ▇
E2E-SWE: Benchmarking LLMs on Building … 1 2% ▇
One Success Isn't Reliability: Thinking… 1 1% ▇
Video2World: Benchmarking Coding Agents… 1 1% ▇
Moonshot AI vendor-reported benchmark s… 1 1% ▇
Miðeind Icelandic LLM Leaderboard 1 1% ▇
Zapier AutomationBench README: public-s… 1 1% ▇
prior: pull toward the middle 0% ▇
weighted average = 75.4
Licences: Epoch AI ([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)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Roblox LLM Leaderboard ([MIT](../../licenses/roblox-llm-leaderboard-MIT.txt)) · GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · τ²/τ³-bench leaderboard: Sierra's own evaluation runs ([MIT](https://opensource.org/license/mit)) · E2E-SWE: Benchmarking LLMs on Building Working Codebases from Scratch ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · One Success Isn't Reliability: Thinkingbox ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Video2World: Benchmarking Coding Agents for Interactive World Modeling from Embodied Videos ([CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/)) · Miðeind Icelandic LLM Leaderboard ([CC BY 4.0](../../licenses/mideind-icelandic-llm-leaderboard-CC-BY-4.0.txt)) · Zapier AutomationBench README: public-set pass rates ([MIT](../../licenses/zapier-automationbench-readme-MIT.txt))
## How to run it
### Through an API
api (Moonshot AI) · $6.00/M blended · $3 in / $15 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-k3.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-k3.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 K3 (ud-q4_k_xl, needs ~1680.5 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/Kimi-K3-GGUF/tree/main))
```sh
llama cli -hf unsloth/Kimi-K3-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-K3-GGUF:UD-Q4_K_XL
```
Other versions: ud-q2_k_xl (~840.5 GB) `llama cli -hf unsloth/Kimi-K3-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 K3 (ud-q4_k_xl, needs ~1680.5 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/Kimi-K3-GGUF/tree/main))
```sh
lms get https://huggingface.co/unsloth/Kimi-K3-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 (~840.5 GB) `lms get https://huggingface.co/unsloth/Kimi-K3-GGUF@UD-Q2_K_XL -y`
Model repository and documentation: [huggingface.co/moonshotai](https://huggingface.co/moonshotai/Kimi-K3)
[interactive version](https://whatmodel.sh/model/kimi-k3) · [JSON](https://whatmodel.sh/api?format=json&all=1)