# Kimi K3Max > Source: Benchize (https://benchize.com) — https://benchize.com/models/kimi-k3-max. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/models/kimi-k3-max. MoonshotAI Canonical URL: https://benchize.com/models/kimi-k3-max Last updated: Wed, 21 Oct 2026 18:52:20 GMT Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations. MoonshotAI ## Benchmarks | Benchmark | Score | ± | Rank | Cost | Latency | Source | | | --- | --- | --- | --- | --- | --- | --- | --- | | [DeepSearchQA (f1-score)](https://benchize.com/benchmarks/deepsearchqa.md) | 95% | — | 1 / 3 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `f1` \| 95.0 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/10/scores/0` | | | | | | | | | [GDPval-AA v2 (Elo-score)](https://benchize.com/benchmarks/gdpval-aa-v2.md) | 1,668elo | — | 3 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1668 \| elo \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/8/scores/0` | | | | | | | | | [BabyVision w/ python](https://benchize.com/benchmarks/babyvision-python.md) | 85.7% | — | 3 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 85.7 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 Temperature 1 thinkingEnabled true tools ["python"] toolsEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/29/scores/0` | | | | | | | | | [BrowseComp](https://benchize.com/benchmarks/browsecomp.md) | 91.2% | — | 1 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 91.2 \| percent \| reported \| Higher is better \| ### Configuration contextCompactionTokens 300000 Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/9/scores/0` | | | | | | | | | [SWE Marathon](https://benchize.com/benchmarks/swe-marathon.md) | 42% | — | 1 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 42.0 \| percent \| reported \| Higher is better \| ### Configuration datasetAdaptation H20-calibrated harness Claude Code Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/4/scores/0` | | | | | | | | | [GPQA Diamond](https://benchize.com/benchmarks/gpqa-diamond.md) | 93.5% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 93.5 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/20/scores/0` | | | | | | | | | [ZeroBench main (pass@5)](https://benchize.com/benchmarks/zero-bench-main.md) | 23% | — | 1 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_at_5` \| 23.0 \| percent \| main_pass_at_5 \| Higher is better \| ### Configuration Reasoning effort max runs 5 subset main Temperature 1 thinkingEnabled true toolsEnabled false topP 1 trials 5 ### Reported statistics runCount 5 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/30/scores/0` | | | | | | | | | [Job Bench](https://benchize.com/benchmarks/jobbench.md) | 52.9% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 52.9 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/14/scores/0` | | | | | | | | | [MLS Bench](https://benchize.com/benchmarks/mls-bench-lite.md) | 48.3% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 48.3 \| percent \| reported \| Higher is better \| ### Configuration harness Kimi Code Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/6/scores/0` | | | | | | | | | [ProgramBench / overall](https://benchize.com/benchmarks/vals-ai%3Aprogrambench%3Aoverall.md) | 77.8% | — | 1 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 77.8 \| percent \| reported \| Higher is better \| ### Configuration harness Kimi Code Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/1/scores/0` | | | | | | | | | [MMMU-Pro w/ python](https://benchize.com/benchmarks/mmmu-pro-python.md) | 83.4% | — | 3 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 83.4 \| percent \| reported \| Higher is better \| ### Configuration imageOrder original_images_prepended Reasoning effort max runs 3 Temperature 1 thinkingEnabled true tools ["python"] toolsEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/24/scores/0` | | | | | | | | | [Toolathlon-Verified](https://benchize.com/benchmarks/toolathlon-verified.md) | 73.2% | — | 5 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 73.2 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/11/scores/0` | | | | | | | | | [OfficeQA Pro — Agent Harness](https://benchize.com/benchmarks/officeqa-pro%3Aagent.md) | 63.3% | — | 3 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 63.3 \| percent \| reported \| Higher is better \| ### Configuration corpusFormat rendered_pdf_images harness Claude Code machineReadablePdfText false Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/17/scores/0` | | | | | | | | | [WorldVQA ForceAnswer](https://benchize.com/benchmarks/worldvqa-force-answer.md) | 51% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 51.0 \| percent \| reported \| Higher is better \| ### Configuration forceAnswer true Reasoning effort max runs 3 Temperature 1 thinkingEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/32/scores/0` | | | | | | | | | [APEX-Agents](https://benchize.com/benchmarks/apex-agents.md) | 41% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 41.0 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/16/scores/0` | | | | | | | | | [MathVision](https://benchize.com/benchmarks/mathvision.md) | 94.3% | — | 3 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 94.3 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 Temperature 1 thinkingEnabled true toolsEnabled false topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/27/scores/0` | | | | | | | | | [MCP Atlas](https://benchize.com/benchmarks/scale%3Amcp_atlas.md) | 84.2% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 84.2 \| percent \| reported \| Higher is better \| ### Configuration judgeModel Gemini 3.1 Pro maxTurns 100 Reasoning effort max subset public taskCount 500 Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/12/scores/0` | | | | | | | | | [MMMU Pro / overall](https://benchize.com/benchmarks/vals-ai%3Ammmu%3Aoverall.md) | 81.6% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 81.6 \| percent \| reported \| Higher is better \| ### Configuration imageOrder original_images_prepended Reasoning effort max runs 3 Temperature 1 thinkingEnabled true toolsEnabled false topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/23/scores/0` | | | | | | | | | [DeepSWE](https://benchize.com/benchmarks/deepswe%3Av1.1.md) | 67.5% | — | 3 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 67.5 \| percent \| reported \| Higher is better \| ### Configuration harness Kimi Code Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/0/scores/0` | | | | | | | | | [DECK-Bench (Internal)](https://benchize.com/benchmarks/deck-bench.md) | 73.5% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 73.5 \| percent \| reported \| Higher is better \| ### Configuration benchmarkVisibility internal Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/19/scores/0` | | | | | | | | | [CharXiv (RQ) w/ python](https://benchize.com/benchmarks/charxiv-rq-python.md) | 91.3% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 91.3 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 subset RQ Temperature 1 thinkingEnabled true tools ["python"] toolsEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/26/scores/0` | | | | | | | | | [DeepSWE](https://benchize.com/benchmarks/deepswe%3Av1.1.md) | 68.51% | Range 63.98–73.05% | 19 / 70 | $4.65 | 4,541.45s | [deepswe](https://deepswe.datacurve.ai/artifacts/v1.1/leaderboard-live.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_at_1` \| 0.6851441241685144 \| score_fraction \| scored_rollout_attempts \| Higher is better \| \| `pass_at_4` \| 0.8938053097345132 \| score_fraction \| tasks_with_one_passing_rollout \| Higher is better \| \| `cost_per_attempt` \| 4.654682129933482 \| USD \| scored_attempt \| Lower is better \| \| `duration` \| 4541.447376412417 \| seconds \| scored_attempt \| Lower is better \| ### Configuration harness mini-swe-agent Reasoning effort max sourceConfiguration mini_swe_agent_kimi_k3_max ### Reported statistics Uncertainty Reported interval: 63.98–73.05% interval [0.6397739833756912,0.7305142649613376] intervalMethod 95% run-to-run: SE across repeated whole-benchmark passes (1.96 * std(runs)/sqrt(R)) runCount 4 sampleSize 451 taskCount 113 ### Source [Original source](https://deepswe.datacurve.ai/artifacts/v1.1/leaderboard-live.json) Source deepswe Original model label kimi-k3 Observed 2026-10-21T16:30:23.371Z Evidence SHA-256 `a7c15d66288fd249c020b9931c017b92d1a3b90e480b3ff34974b752bd030019` Source locator `/rows/18` | | | | | | | | | [Hyper-τ-bench](https://benchize.com/benchmarks/sierra%3Ahyper-tau.md) | 16.1% | — | 5 / 6 | — | — | [sierra](https://raw.githubusercontent.com/sierra-research/hyper-tau-bench/main/web/leaderboard/public/hyper-submissions/kimi-code_kimi-k3/submission.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `overall` \| 16.1 \| percent \| mean_reward \| Higher is better \| \| `airline` \| 42.3 \| percent \| mean_reward \| Higher is better \| \| `retail` \| 49.1 \| percent \| mean_reward \| Higher is better \| \| `telecom` \| 24.9 \| percent \| mean_reward \| Higher is better \| \| `banking` \| 4.4 \| percent \| mean_reward \| Higher is better \| ### Configuration harness Kimi Code methodology Not reported modality developer_agent Reasoning effort max sourceSubmission kimi-code_kimi-k3 ### Reported statistics baseline true ### Source [Original source](https://raw.githubusercontent.com/sierra-research/hyper-tau-bench/main/web/leaderboard/public/hyper-submissions/kimi-code_kimi-k3/submission.json) Source sierra Original model label Kimi K3 Evaluated 2026-08-31 Observed 2026-10-21T16:55:06.483Z Evidence SHA-256 `e5c2fe1cfe6468373e43c54594a581f85126f1bd002bb9b76d53f8df5d68c809` Source locator `/rows/0` | | | | | | | | | [HLE-Full w/ tools](https://benchize.com/benchmarks/hle-full-tools.md) | 56% | — | 4 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 56.0 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true toolsEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/22/scores/0` | | | | | | | | | [HLE-Full](https://benchize.com/benchmarks/hle-full.md) | 43.5% | — | 4 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 43.5 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true toolsEnabled false topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/21/scores/0` | | | | | | | | | [OmniDocBench](https://benchize.com/benchmarks/omnidocbench.md) | 91.1% | — | 1 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 91.1 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 Temperature 1 thinkingEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/33/scores/0` | | | | | | | | | [ZeroBench main w/ python (pass@5)](https://benchize.com/benchmarks/zero-bench-main-python.md) | 41% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `pass_at_5` \| 41.0 \| percent \| main_pass_at_5 \| Higher is better \| ### Configuration Reasoning effort max runs 5 subset main Temperature 1 thinkingEnabled true tools ["python"] toolsEnabled true topP 1 trials 5 ### Reported statistics runCount 5 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/31/scores/0` | | | | | | | | | [AutomationBench](https://benchize.com/benchmarks/zapier%3Aautomationbench.md) | 30.8% | — | 1 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 30.8 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max subset public taskCount 600 Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/13/scores/0` | | | | | | | | | [SpreadsheetBench v2-full](https://benchize.com/benchmarks/spreadsheetbench%3Av2-full.md) | 34.8% | — | 1 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 34.8 \| percent \| reported \| Higher is better \| ### Configuration harness Claude Code Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/18/scores/0` | | | | | | | | | [Kimi Code Bench 2.0 (Internal)](https://benchize.com/benchmarks/kimi-code-bench-2.md) | 72.9% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 72.9 \| percent \| reported \| Higher is better \| ### Configuration benchmarkVisibility internal Reasoning effort max reportedHarnesses ["Kimi Code","Claude Code"] Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/7/scores/0` | | | | | | | | | [Terminal-Bench 2.1](https://benchize.com/benchmarks/artificial-analysis%3Aterminalbench_v2_1.md) | 88.3% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 88.3 \| percent \| reported \| Higher is better \| ### Configuration harness Kimi Code Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/2/scores/0` | | | | | | | | | [Hyper-τ-bench](https://benchize.com/benchmarks/sierra%3Ahyper-tau.md) | 17.9% | — | 4 / 6 | — | — | [sierra](https://raw.githubusercontent.com/sierra-research/hyper-tau-bench/main/web/leaderboard/public/hyper-submissions/opencode_kimi-k3/submission.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `overall` \| 17.9 \| percent \| mean_reward \| Higher is better \| \| `airline` \| 43.8 \| percent \| mean_reward \| Higher is better \| \| `retail` \| 52.1 \| percent \| mean_reward \| Higher is better \| \| `telecom` \| 27.6 \| percent \| mean_reward \| Higher is better \| \| `banking` \| 6 \| percent \| mean_reward \| Higher is better \| ### Configuration harness OpenCode methodology Not reported modality developer_agent Reasoning effort max sourceSubmission opencode_kimi-k3 ### Reported statistics baseline true ### Source [Original source](https://raw.githubusercontent.com/sierra-research/hyper-tau-bench/main/web/leaderboard/public/hyper-submissions/opencode_kimi-k3/submission.json) Source sierra Original model label Kimi K3 Evaluated 2026-09-02 Observed 2026-10-21T16:55:06.818Z Evidence SHA-256 `a42ce6d899adc5547595c8f65c2ac2b95835a8cd04280522fe8d808b9adeda0c` Source locator `/rows/0` | | | | | | | | | [MathVision w/ python](https://benchize.com/benchmarks/mathvision-python.md) | 97.8% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 97.8 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 Temperature 1 thinkingEnabled true tools ["python"] toolsEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/28/scores/0` | | | | | | | | | [ToneBench Scriptwriting](https://benchize.com/benchmarks/tonebench%3Ascripts.md) | 2,176.7 | Range 2,128.6–2,216.8 | 20 / 126 | — | — | [tonebench](https://benchmark.towardsai.com/data/leaderboard.json) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 2176.7 \| rating \| reported \| Higher is better \| \| `overall` \| 82.23 \| percent \| reported \| Higher is better \| \| `anti_slop_numbers` \| 70.94 \| source_score \| rubric \| Higher is better \| \| `continuity_emotion` \| 82.28 \| source_score \| rubric \| Higher is better \| \| `cue_quality` \| 79.98 \| source_score \| rubric \| Higher is better \| \| `hook_strength` \| 83.13 \| source_score \| rubric \| Higher is better \| \| `length_adherence` \| 94.23 \| source_score \| rubric \| Higher is better \| \| `slop` \| 83.44 \| source_score \| rubric \| Higher is better \| \| `slop_score` \| 95.94 \| source_score \| rubric \| Higher is better \| \| `substance_accuracy` \| 83.54 \| source_score \| rubric \| Higher is better \| \| `tone_voice` \| 78.73 \| source_score \| rubric \| Higher is better \| \| `writing_quality` \| 78.83 \| source_score \| rubric \| Higher is better \| \| `youtube_best_practices` \| 81.24 \| source_score \| rubric \| Higher is better \| ### Configuration provider openrouter Reasoning effort unknown scoreScope 10_task_current sourceConfiguration kimi-k3-or thinkingEnabled true ### Reported statistics Uncertainty Reported interval: 2,128.6–2,216.8 interval [2128.6,2216.8] rank 20 runCount 5 ### Source [Original source](https://benchmark.towardsai.com/data/leaderboard.json) Source tonebench Original model label Kimi K3 (max) Evaluated 2026-10-08 Observed 2026-10-21T16:32:48.862Z Evidence SHA-256 `4264338337f418d3ded7ace250cad64fe12cacbd4b79a312bfaf0ae9268a01c3` Source locator `/rows/19` | | | | | | | | | [FrontierSWE](https://benchize.com/benchmarks/frontierswe%3Adominance.md) | 81.2% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `dominance` \| 81.2 \| percent \| dominance_2026-07-16 \| Higher is better \| ### Configuration harness Kimi Code normalization dominance normalizationDate 2026-07-16 Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/3/scores/0` | | | | | | | | | [AA-Briefcase (Elo-score)](https://benchize.com/benchmarks/aa-briefcase.md) | 1,548elo | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1548 \| elo \| reported \| Higher is better \| ### Configuration Reasoning effort max Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/15/scores/0` | | | | | | | | | [PerceptionBench](https://benchize.com/benchmarks/perceptionbench.md) | 58.5% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 58.5 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 Temperature 1 thinkingEnabled true topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/34/scores/0` | | | | | | | | | [CharXiv (RQ)](https://benchize.com/benchmarks/charxiv-rq.md) | 84.8% | — | 2 / 5 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 84.8 \| percent \| reported \| Higher is better \| ### Configuration Reasoning effort max runs 3 subset RQ Temperature 1 thinkingEnabled true toolsEnabled false topP 1 ### Reported statistics runCount 3 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/25/scores/0` | | | | | | | | | [PostTrain Bench](https://benchize.com/benchmarks/posttrainbench.md) | 36.6% | — | 2 / 6 | — | — | [Moonshot AI · sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c](https://www.kimi.ai/blog/kimi-k3) | | | ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `score` \| 36.6 \| percent \| reported \| Higher is better \| ### Configuration gpu H20 harness Claude Code Reasoning effort max runs 3 Temperature 1 thinkingEnabled true topP 1 ### Source Score publisher [Moonshot AI](https://www.kimi.ai/blog/kimi-k3) Collected via [moonshot k3](https://www.kimi.ai/blog/kimi-k3) Publisher reference Referenced by the collection source; the upstream result has not been independently captured. Benchmark [https://www.kimi.ai/blog/kimi-k3](https://www.kimi.ai/blog/kimi-k3) [Collected via](https://www.kimi.ai/blog/kimi-k3) Source Moonshot AI Original model label Kimi K3 (max) Source version sha256:0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c Observed 2026-10-21T16:31:44.130Z Evidence SHA-256 `0e806bbad76dc8e8dbc899a86051e3406bdd5995fe6c6aa1b2176dde4fb28e6c` Source locator `/flight/sheet/data/5/scores/0` | | | | | | | | --- Cite as: Kimi K3Max — Benchize, https://benchize.com/models/kimi-k3-max (retrieved 2026-10-21)