Kimi K3
Made by Moonshot, released 16 Jul 2026. Epoch AI lists its access as “Open weights (non-commercial)”. Appears strongest at coding among the models listed here.
In short
What it does well
- Coding: Top tier (14th of 77 listed models)
- Knowledge: Strong (22nd of 43 listed models)
- Maths: Strong (19th of 43 listed models)
- Writing: Strong (23rd of 77 listed models)
Where it falls short
The four main ratings show no weakness. That is not evidence it has none.
What we don't know yet
- Fixing real software bugs: Not enough data
The four ratings
- Top tier
- Strong
- Middle
- Behind the leaders
- Not enough data
Ratings compare only the 77 models listed here, in quarters, best first. “Not enough data” means no result yet. It is never a low score.
- Writing
- StrongHow often people preferred its answers to writing prompts. 1,457 on Arena's scale · Source for Kimi K3 writing score (lmarena)
- Coding
- Top tierHow often people preferred its answers to coding questions. 1,508 on Arena's scale · Source for Kimi K3 coding score (lmarena)
- Maths
- StrongHow many very hard maths problems it solved. 72.2% · Source for Kimi K3 maths score (epoch)
- Knowledge
- StrongHow many short factual questions it answered correctly. 50.6% · Source for Kimi K3 knowledge score (epoch)
Evidence completeness: Broad evidence, 12 of 13 tests have a result.
Where it is relatively stronger or weaker
This compares each kind of request with the model's own overall position among the 77 listed models, using Arena's votes. It is about performance on different languages, kinds of request and topics, not about political or social bias.
Relatively stronger
No category stands out as clearly stronger than its overall position.
Relatively weaker
No category stands out as clearly weaker than its overall position.
Show every category we can see
| Category | Position | Arena score | Votes | Source |
|---|---|---|---|---|
| Expert-level prompts Kind of request | 10th of 77 | 1,520 95% range 1,509 to 1,530 | 3,347 | Source for Expert-level prompts |
| Hard prompts Kind of request | 14th of 77 | 1,496 95% range 1,490 to 1,501 | 23,477 | Source for Hard prompts |
| Long prompts Kind of request | 13th of 77 | 1,493 95% range 1,487 to 1,499 | 15,841 | Source for Long prompts |
| Longer conversations Kind of request | 12th of 77 | 1,490 95% range 1,481 to 1,498 | 5,628 | Source for Longer conversations |
| Chinese Language | 16th of 76 | 1,531 95% range 1,519 to 1,542 | 2,688 | Source for Chinese |
| English prompts Language | 18th of 77 | 1,481 95% range 1,475 to 1,487 | 13,907 | Source for English prompts |
| French Language | 26th of 72 | 1,486 95% range 1,469 to 1,503 | 1,434 | Source for French |
| German Language | 13th of 55 | 1,489 95% range 1,467 to 1,512 | 712 | Source for German |
| Japanese Language | 8th of 54 | 1,495 95% range 1,471 to 1,519 | 725 | Source for Japanese |
| Korean Language | 8th of 54 | 1,459 95% range 1,437 to 1,480 | 769 | Source for Korean |
| Non-English prompts Language | 22nd of 77 | 1,466 95% range 1,461 to 1,471 | 22,861 | Source for Non-English prompts |
| Russian Language | 16th of 77 | 1,484 95% range 1,475 to 1,493 | 4,663 | Source for Russian |
| Spanish Language | 16th of 70 | 1,477 95% range 1,459 to 1,495 | 1,158 | Source for Spanish |
| Entertainment, sport and media Topic | 20th of 77 | 1,452 95% range 1,445 to 1,459 | 9,062 | Source for Entertainment, sport and media |
| Law and government Topic | 9th of 76 | 1,502 95% range 1,490 to 1,513 | 2,859 | Source for Law and government |
| Medicine and health Topic | 15th of 75 | 1,487 95% range 1,475 to 1,500 | 2,489 | Source for Medicine and health |
| Science Topic | 9th of 77 | 1,508 95% range 1,499 to 1,516 | 5,732 | Source for Science |
| Software and IT Topic | 16th of 77 | 1,499 95% range 1,493 to 1,505 | 13,902 | Source for Software and IT |
Bias: what we can and can't see
Our sources do not measure political lean, stereotyping, flattery (agreeing with you too readily) or safety behaviour, so we cannot say how any model does on those. What we can show is above: where a model is relatively stronger or weaker across languages, kinds of request and topics. Also, people voting in Arena can favour answer style, such as length and formatting, and our numbers do not adjust for that. See what AI bias means and what these tests can tell you.
Best-supported use cases
- Homework and maths: best open-weights option.
- Business emails and everyday work: best open-weights option.
Where an alternative might be more suitable
- Writing and creating content: rated Strong on writing votes. Compare it with Claude Opus 5.5, the best-supported choice for this task.
- Homework and maths: rated Strong on hard maths problems. Compare it with GPT-6.1 Sol, the best-supported choice for this task.
- Research and checking facts: rated Strong on short factual questions. Compare it with Gemini 4 Argon, the best-supported choice for this task.
What it is
- Maker
- Moonshot
- Release date (Epoch AI)
- 16 Jul 2026
- Access, as listed by Epoch AI
- Open weights (non-commercial)
- Licence, as listed by Arena
- Kimi K3 license
- Maker's country (Epoch AI)
- China
- Last observation in our data
- 28 Oct 2026
- Data last changed
- 29 Oct 2026, 14:36 UTC
Epoch lists access as 'Open weights (non-commercial)'. A newer 'Kimi K3.1 Pro' is a different model and is not mapped here.
We have not verified an official product page for this model, so we do not link to one.
How you can use it
Open weights. Arena lists its licence as Kimi K3 license. In plain terms, the model's files are published under a licence that lets others run it, so it can be offered by more than one company, sometimes at lower cost. That does not guarantee that any particular app is free. Open licences differ, and some restrict commercial use, so check the licence terms before building a business on it.
How you can reach it (an app, a website, a developer API or a third-party tool) and what that costs are not covered by our benchmark data. Where we could verify prices from the maker's own pages, they are shown under Price. Epoch AI lists access as “Open weights (non-commercial)”. See models, apps and harnesses and how AI pricing works.
Price
Price not independently verified
We have not been able to check this maker's own pricing page, so we show nothing for it.
We never infer subscription prices from API prices. New to “tokens” and “per million”? See how AI pricing works.
Good to remember
- Benchmarks are samples of tasks. None proves performance in every situation.
- Arena scores come from people choosing between answers. That measures preference, not correctness.
- Ratings are relative to the 77 models listed here. A different set would change them.
- Sources test models at different settings. Each result shows the setting used, and we do not adjust it.
The evidence, for the curious
Show all 13 tests with settings and sources
| Test | Result | Rating | Setting used | Source identifier | Observed | Links |
|---|---|---|---|---|---|---|
| Arena text: overall preference Arena 'text' leaderboard (Bradley-Terry rating, no style control) |
1,476 Arena score 95% range 1,471 to 1,480, 36,780 votes |
Top tier | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · text/latest |
28 Oct 2026 | Source for Kimi K3 Arena text: overall preference score (lmarena) |
| Arena text: creative writing Arena 'text' leaderboard, category creative_writing |
1,457 Arena score 95% range 1,449 to 1,465, 7,452 votes |
Strong | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · text/latest |
28 Oct 2026 | Source for Kimi K3 Arena text: creative writing score (lmarena) |
| Arena text: coding Arena 'text' leaderboard, category coding |
1,508 Arena score 95% range 1,501 to 1,515, 9,235 votes |
Top tier | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · text/latest |
28 Oct 2026 | Source for Kimi K3 Arena text: coding score (lmarena) |
| Arena text: maths Arena 'text' leaderboard, category math |
1,492 Arena score 95% range 1,478 to 1,507, 1,673 votes |
Top tier | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · text/latest |
28 Oct 2026 | Source for Kimi K3 Arena text: maths score (lmarena) |
| Arena text: following instructions Arena 'text' leaderboard, category instruction_following |
1,485 Arena score 95% range 1,479 to 1,492, 12,229 votes |
Top tier | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · text/latest |
28 Oct 2026 | Source for Kimi K3 Arena text: following instructions score (lmarena) |
| Arena text: business and management Arena 'text' leaderboard, category industry_business_and_management_and_financial_operations |
1,469 Arena score 95% range 1,461 to 1,477, 6,887 votes |
Top tier | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · text/latest |
28 Oct 2026 | Source for Kimi K3 Arena text: business and management score (lmarena) |
| Arena WebDev: building web apps Arena 'webdev' leaderboard (Code Arena), category overall |
1,656 Arena score 95% range 1,651 to 1,662, 21,281 votes |
Top tier | effort: Max, bradley-terry rating | Kimi K3 (Max) lmarena · webdev/latest |
28 Oct 2026 | Source for Kimi K3 Arena WebDev: building web apps score (lmarena) |
| FrontierMath Tiers 1-3 (v2) FrontierMath Tiers 1-3 v2 (private set), benchmark released 2026-06-12 |
72.2% share of problems solved standard error ±2.7 points |
Strong | effort: max, best score across scorers | kimi-k3_max epoch · frontiermath_tiers_1_3_v2.csv |
17 Jul 2026 | Source for Kimi K3 FrontierMath Tiers 1-3 (v2) score (epoch) No public run log in the data file. |
| OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025, benchmark released 2024-12-19 |
97.2% share of problems solved standard error ±1.1 points |
Strong | effort: max, best score across scorers | kimi-k3_max epoch · otis_mock_aime_2024_2025.csv |
16 Jul 2026 | Source for Kimi K3 OTIS Mock AIME 2024-2025 score (epoch) No public run log in the data file. |
| GPQA Diamond GPQA Diamond, benchmark released 2023-11-20 |
93.1% share of questions correct standard error ±1.5 points |
Strong | effort: max, best score across scorers | kimi-k3_max epoch · gpqa_diamond.csv |
16 Jul 2026 | Source for Kimi K3 GPQA Diamond score (epoch) No public run log in the data file. |
| SimpleQA Verified SimpleQA Verified, benchmark released 2025-09-09 |
50.6% share of questions correct standard error ±1.6 points |
Strong | effort: max, best score across scorers | kimi-k3_max epoch · simpleqa_verified.csv |
27 Aug 2026 | Source for Kimi K3 SimpleQA Verified score (epoch) · Run details for SimpleQA Verified |
| SWE-bench Verified (Epoch runs) | Not enough data | |||||
| FrontierCode (reported by Cognition) FrontierCode, benchmark released 2026-06-08, aggregation Mean@5 |
44.2% mean score over 5 attempts no uncertainty figure published |
Not rated (results use different tools) | effort: none, tool: mini-swe-agent, mean@5 | kimi-k3_unknown epoch · frontiercode_external.csv |
not provided | Source for Kimi K3 FrontierCode (reported by Cognition) score (epoch) · Run details for FrontierCode (reported by Cognition) |
Missing evidence
Not enough data for these tests:
- SWE-bench Verified (Epoch runs)
Missing results are not low scores and do not count against the model.
Where a source gives no public run log, we cannot show the run behind the score. See data sources.
Not sure this is the right one?
Tell us what you need help with and we'll narrow the options down.