Identity
- Type
- startup
- Country label
- China
- HQ country
- China
- Founding country
- China
- Parent
- not in the notes
- Parent country
- not in the notes
- Country basis
- the research notes
Where this comes from
Convention D1: a dual label, HQ first, when the HQ, founding or parent country differ.
Models by Class
Models
ModelClassOpennessParametersReleased
Kimi K3multimodal LLM16 Jul 2026Kimi K2.7 CodecodeJun 2026Kimi K2.6LLMApr 2026Kimi K2.5multimodal LLMJan 2026Kimi K2LLMJul 2025Kimi-VLmultimodal LLM9 Apr 2025Kimi-VL-A3B-Instructmultimodal LLM9 Apr 2025Kimi-VL-A3B-Thinkingmultimodal LLM9 Apr 2025Kimi-Audiospeech recognition2025KimiLLM–Kimi K2-0905LLM–Kimi K2 Thinkingreasoning–Links
OrganisationMoonshot AIstartup, China
Moonshot AIstartup, ChinaModelsKimi K3Moonshot AI, multimodal LLMKimi K2.7 CodeMoonshot AI, codeKimi K2.6Moonshot AI, LLMKimi K2.5Moonshot AI, multimodal LLM+8 more
Kimi K3Moonshot AI, multimodal LLMKimi K2.7 CodeMoonshot AI, codeKimi K2.6Moonshot AI, LLMKimi K2.5Moonshot AI, multimodal LLM+8 moreUse Casesnone recorded
Who Runs Itnone recorded
Models 12
KimiMoonshot AI, LLMDeveloperKimi K2.5Moonshot AI, multimodal LLMDeveloperKimi K2.6Moonshot AI, LLMDeveloperKimi K3Moonshot AI, multimodal LLMDeveloperKimi-VL-A3B-ThinkingMoonshot AI, multimodal LLMDeveloperKimi K2 ThinkingMoonshot AI, reasoningDeveloperKimi K2-0905Moonshot AI, LLMDeveloperKimi-VLMoonshot AI, multimodal LLMDeveloperKimi-VL-A3B-InstructMoonshot AI, multimodal LLMDeveloperKimi K2Moonshot AI, LLMDeveloperKimi K2.7 CodeMoonshot AI, codeDeveloperKimi-AudioMoonshot AI, speech recognitionDeveloperScores
| Benchmark | Model | Score | Who ran it | Setting |
|---|---|---|---|---|
| Arena Text leaderboard | Kimi K2.6 | 1,461 Elo | independent | |
| Arena Text leaderboard | Kimi K3 | 1,488 Elo | independent | kimi-k3-max; top open model |
| Artificial Analysis Intelligence Index | Kimi K3 | 44 index | independent | Kimi K3 (Max) |
| Artificial Analysis Intelligence Index | Kimi K3 | 44 index | independent | max |
| BrowseComp | Kimi K3 | 91.2 % | vendor-reported | |
| DeepSWE (version not stated) | Kimi K3 | 67.5 % | vendor-reported | Kimi Code harness; 67.3 with mini-SWE-agent |
| FrontierSWE | Kimi K3 | 81.2 | vendor-reported | Kimi Code harness |
| GPQA Diamond | Kimi K3 | 93.5 % | vendor-reported | |
| Humanity's Last Exam (HLE) | Kimi K3 | 43.5 % | vendor-reported | without tools |
| Humanity's Last Exam (HLE) | Kimi K3 | 56 % | vendor-reported | with tools |
| Kimi Code Bench | Kimi K2.7 Code | 62 | vendor-reported | |
| OSWorld-Verified | Kimi K3 | 84.8 % | vendor-reported | |
| ProgramBench | Kimi K2.7 Code | 53.6 % | vendor-reported | Kimi Code CLI, thinking |
| ProgramBench | Kimi K3 | 77.8 % | vendor-reported | Kimi Code harness |
| SWE-bench Verified | Kimi K2.5 | 76.8 % | vendor-reported | |
| SWE-bench Verified | Kimi K2.6 | 80.2 % | vendor-reported | |
| SWE-Marathon | Kimi K3 | 42 | vendor-reported | Claude Code harness |
| Terminal-Bench | Kimi K3 | 88.3 % | vendor-reported | |
| Terminal-Bench | Kimi K3 | 88.3 % | vendor-reported | Kimi Code harness |
Where this comes from
Every recorded score of this organisation's models, with who ran it. Scores from different settings are not directly comparable.
Sources
- Kimi K3 (max) - Intelligence, Performance & Price Analysis | Artificial Analysis artificialanalysis.ai https://artificialanalysis.ai/models/kimi-k3
- huggingface.co/moonshotai https://huggingface.co/moonshotai
- Speech Recognition Accuracy Leaders, 2022–2026 select.keenable.ai https://select.keenable.ai/r/speech-recognition-accuracy-leaders-2022-2026/JbURQjlVooLQ6J10NkbRVw
- www-web.itiger.com/news/1119147261 https://www-web.itiger.com/news/1119147261
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Moonshot AI", https://aurion.semai.id/organisations/moonshot-ai/, records of October 2026.