Updated daily at 05:00 UTC
- Released
- 16 Jul 2026
- Status
- Current
- Context
- 1M
- Max output
- —
- Input per 1M tokens
- $3.00
- Output per 1M tokens
- $15.00
- Open weights
- Yes
- License
- kimi-k3
- Parameters
- 2800B, 104B active
- Knowledge cutoff
- —
- Reasoning
- Yes
- Modalities
- text, image → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
| Category | Score | Rank | Coverage |
|---|---|---|---|
| Knowledge & reasoning | 50.2 | 15 | 3 of 4 |
| Coding & software engineering | 56.3 | Not ranked | 2 of 5 |
| Factuality & hallucination | 50.6 | 21 | 1 of 1 |
| Human preference | 72.0 | Not ranked | 2 of 5 |
Benchmark results
Every result we hold for this model, with who ran it and where it was published.
Knowledge & reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Chess Puzzles | Puzzles solved | 39.0% | Independent | max | 16 Jul 2026 | Epoch AI |
| Chess Puzzles | Puzzles solved | 25.0% | Independent | high | 7 Aug 2026 | Epoch AI |
| Chess Puzzles | Puzzles solved | 20.0% | Independent | low | 7 Aug 2026 | Epoch AI |
| DTBench (Decision Theory Benchmark) | Accuracy (capability questions) | 91.2% | Third party | max | — | https://conceptualreasoning.ai/dtbench via Epoch AI |
| Humanity's Last Exam (HLE) | Accuracy | 56.0% | Self-reported | — | — | Model Card via Hugging Face Community Evals |
| LMCA (Language Model Conceptual Argumentation) | Correlation with expert ratings (Pearson r) | 0.5274 | Third party | max | — | https://conceptualreasoning.ai/lmca via Epoch AI |
| Mystery Game Puzzles | Puzzles solved | 26.0% | Independent | max | 29 Jul 2026 | Epoch AI |
| Mystery Game Puzzles | Puzzles solved | 12.0% | Independent | low | 29 Aug 2026 | Epoch AI |
| SimpleBench | Score (AVG@5) | 60.7% | Third party | max | — | https://lmcouncil.ai/benchmarks via Epoch AI |
Scientific reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GPQA Diamond | Accuracy | 93.1% | Independent | max | 16 Jul 2026 | Epoch AI |
| GPQA Diamond | Accuracy | 91.9% | Independent | high | 7 Aug 2026 | Epoch AI |
| GPQA Diamond | Accuracy | 84.8% | Independent | low | 7 Aug 2026 | Epoch AI |
| GPQA Diamond | Accuracy | 93.5% | Self-reported | — | — | Model Card via Hugging Face Community Evals |
| Surface Evolver Bench | Mean score (partial credit) | 95.0% | Third party | unknown | — | https://yhenon.github.io/surface-evolver-llm-eval/ via Epoch AI |
| Surface Evolver Bench | Mean score (partial credit) | 93.0% | Third party | max | — | https://yhenon.github.io/surface-evolver-llm-eval/ via Epoch AI |
Mathematics
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| FrontierMath Tier 4 | Accuracy | 39.0% | Independent | max | 17 Jul 2026 | Epoch AI |
| FrontierMath Tiers 1-3 | Accuracy | 72.2% | Independent | max | 17 Jul 2026 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 97.2% | Independent | max | 16 Jul 2026 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 93.3% | Independent | high | 7 Aug 2026 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 68.9% | Independent | low | 7 Aug 2026 | Epoch AI |
Coding & software engineering
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| DeepSWE | Pass@1 | 68.5% | Third party | max | — | https://deepswe.datacurve.ai/ via Epoch AI |
| FrontierCode | Main score (Mean@5) | 44.2% | Third party | unknown | — | https://cognition.com/frontiercode via Epoch AI |
| PostTrainBench | Average score | 32.0% | Source not stated | unknown | — | Epoch AI via Epoch AI |
| WeirdML (v2) | Average accuracy | 82.6% | Third party | max | — | https://htihle.github.io/weirdml.html via Epoch AI |
Professional knowledge work
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| APEX-Agents | Pass@1 (all rubric criteria met) | 50.6% | Source not stated | unknown | — | Epoch AI via Epoch AI |
Vision & multimodal understanding
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Furniture Assembly | Score | 34.2% | Independent | max | 11 Sep 2026 | Epoch AI |
Factuality & hallucination
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| SimpleQA Verified | Accuracy (% correct) | 50.6% | Independent | max | 27 Aug 2026 | Epoch AI |
Human preference
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Code Arena: WebDev (arena.ai) | Arena score (Elo scale) | 1660 | Independent | max | 25 Sep 2026 | Arena |
| Code Arena: WebDev (arena.ai) | Votes | 14739 | Independent | max | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Arena score (Elo scale) | 1488 | Independent | max | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 26400 | Independent | max | 25 Sep 2026 | Arena |
Other Kimi K versions
- Kimi K2.7 Code12 Jun 2026
- Kimi K2.620 Apr 2026
- Kimi K2.527 Jan 2026
- Kimi K2 Thinking6 Nov 2025
- Kimi K211 Jul 2025
- Kimi Thinking PreviewMay 2025
- kimi-latestFeb 2025