Reasoning
198-question Diamond subset; independent AA evaluation
- 01
- 02
- 03
- 03
- 03
- 08
Across the task benchmarks we track, Kimi K3’s strongest relative results are in terminal use, web research, and reasoning. On Terminal-Bench 2.1, it scores 85%, ranking #1 among 13 models with reported results in our dataset.
Updated: Data as of:
Strongest results
| Benchmark | Measures | Score |
|---|---|---|
| Agent Index ↗v1.2 Preview | Overall agent capability | 31.3 / 100#17 of 26 |
| BrowseComp ↗Published open-model reports | Web research | 91.2%#2 of 8 |
| Humanity's Last Exam ↗Stable HLE — text-only subset | Expert reasoning | 46.9%#15 of 68 |
| GPQA ↗Diamond / Artificial Analysis | Reasoning | 93.54%Joint #8 of 30 |
Model details
| Model | Kimi K3 |
|---|---|
| Agent Index v1.2 | 31.3 / 100 · Rank 17 · Preview |
| Creator | Moonshot AI (opens in a new tab) |
| Availability | Open source / open weights |
| Context size | 1,048,576 tokens (opens in a new tab) |
| Max output | 1,048,576 tokens (opens in a new tab) |
| Knowledge cutoff | Not reported |
| License | Kimi K3 License · Model weights ↗ (opens in a new tab) |
Pricing and performance
Per provider and reported configuration. Prices are USD per 1M tokens.
| Provider / configuration | Contexttokens | InputUSD / 1M tokens | OutputUSD / 1M tokens | LatencyTTFT · seconds | Speedtokens / sec |
|---|---|---|---|---|---|
| Kimi K3 (max) | 1,000,000 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.421 (opens in a new tab) | 216.139 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.302 (opens in a new tab) | 90.056 (opens in a new tab) |
| Kimi K3 (max) | 204,800 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 0.87 (opens in a new tab) | 115.162 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.224 (opens in a new tab) | 30.426 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 0.961 (opens in a new tab) | 113.916 (opens in a new tab) |
| Kimi K3 (max) | 1,000,000 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.778 (opens in a new tab) | 175.272 (opens in a new tab) |
| Kimi K3 (max) (FAST) | 1,048,576 | $6.00 (opens in a new tab) | $30.00 (opens in a new tab) | 1.659 (opens in a new tab) | 481.694 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 5.977 (opens in a new tab) | 64.916 (opens in a new tab) |
| Kimi K3 (max) (ULTRA CHAT) | 1,048,576 | $5.60 (opens in a new tab) | $28.00 (opens in a new tab) | Not reported | Not reported |
| Kimi K3 (max) | 1,048,576 | $0.65 (opens in a new tab) | $15.00 (opens in a new tab) | 1.133 (opens in a new tab) | 84.476 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 0.933 (opens in a new tab) | 143.314 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.047 (opens in a new tab) | 212.032 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.187 (opens in a new tab) | 165.949 (opens in a new tab) |
| Kimi K3 (max) | 1,048,576 | $3.00 (opens in a new tab) | $15.00 (opens in a new tab) | 1.319 (opens in a new tab) | 43.821 (opens in a new tab) |
Benchmark performance
Kimi K3 in color, alongside leading models. Latest stored reports, best effort per report date.
GPQA Diamond ↗
Coding
113 tasks; mini-swe-agent
- 01
- 01
- 01
- 04
- 05
- 06
Terminal Use
89 tasks; Terminus 2; e2b; three repeats
- 01
- 02
- 03
- 04
- 05
Web Research
Published reports; source settings retained
- 01
- 02
- 03
- 04
- 05
Work Automations
Private held-out set; API mode; 50 steps; strict pass/fail
- 01
- 02
- 03
- 04
- 05
- 09
Knowledge Work
220 tasks; Stirrup agent; pairwise Elo
- 01
- 02
- 03
- 04
- 05
- 18
Reported benchmarks
All published results in our dataset, with versions, evaluation settings, and sources.
| Benchmark | Score | Evaluation setup | Source |
|---|---|---|---|
| AA-Briefcase1.1 / combined EloCombined Elo · Higher is better | 1,498.42 Elo | selection: best reported configuration per release · Kimi K3 (Max) · tools: benchmark-defined | Artificial Analysis (opens in a new tab)independent |
| AutomationBench1.0.6 / Zapier held-out private setStrict task completion · Higher is better | 22.68% | Kimi K3; private held-out 657 tasks; API mode; 50-step cap; strict completion | Zapier ↗ (opens in a new tab)benchmark publisher |
| AutomationBench-AA1.0.6 / AA private 657Objectives completed without guardrail violations · Higher is better | 58.273% | Kimi K3 (Max); Private 657 tasks; 50 turns; objective completion | Artificial Analysis (opens in a new tab)independent |
| AutomationBench-AA1.0.6 / AA private 657Objectives completed without guardrail violations · Higher is better | 49.257% | Kimi K3 (Low); Private 657 tasks; 50 turns; objective completion | Artificial Analysis (opens in a new tab)independent |
| BrowseCompPublished open-model reportsAccuracy · Higher is better | 91.2% | harness comparison: provider-reported; settings differ across models · Max effort; temperature 1; top_p 1; context compaction triggered at 300K tokens. Primary General Agents benchmark image. Separate no-management 1M track scores 90.4. | Moonshot AI ↗ (opens in a new tab)provider reported |
| DeepSWEv1.1Pass@1 · Higher is better | 69% | mini-swe-agent; effort=max; ±5 percentage points | Datacurve ↗ (opens in a new tab)benchmark publisher |
| GDP.pdf100 tasks / All-pass / five attemptsAccuracy · Higher is better | 22% | selection: best reported configuration per release · Kimi K3 (Max) · tools: benchmark-defined | Artificial Analysis (opens in a new tab)independent |
| GDPval-AA2.1Elo · Higher is better | 1,542.76 Elo | Kimi K3 (Max); 220 tasks; Stirrup agent; pairwise Elo | Artificial Analysis (opens in a new tab)independent |
| GDPval-AA2.1Elo · Higher is better | 1,135.77 Elo | Kimi K3 (Low); 220 tasks; Stirrup agent; pairwise Elo | Artificial Analysis (opens in a new tab)independent |
| GPQADiamond / Artificial AnalysisAccuracy · Higher is better | 93.535% | Kimi K3 (Max); 198-question Diamond subset; independent AA evaluation | Artificial Analysis (opens in a new tab)independent |
| GPQADiamond / Artificial AnalysisAccuracy · Higher is better | 84.242% | Kimi K3 (Low); 198-question Diamond subset; independent AA evaluation | Artificial Analysis (opens in a new tab)independent |
| HLE-Diamond1,000 questions / no toolsAccuracy · Higher is better | 26.9% | selection: best reported configuration per release · Kimi K3 (Max) · No tools | Artificial Analysis (opens in a new tab)independent |
| Humanity's Last ExamHLE — source does not specify subsetAccuracy · Higher is better | 59.8% | temperature=1.0; top_p=0.95; max generation 163840 tokens; max context 300000 tokens; context management; GPT-5.6 Luna medium judge. · subset: unspecified · Tools enabled | Z.ai ↗ (opens in a new tab)other · 2026-08-14 |
| Humanity's Last ExamStable HLE — full setAccuracy · Higher is better | 43.5% | Max effort; temperature=1.0; top_p=0.95. · subset: full · No tools | Moonshot AI ↗ (opens in a new tab)provider reported |
| Humanity's Last ExamStable HLE — full setAccuracy · Higher is better | 56% | Max effort; temperature=1.0; top_p=1.0; general tools. · subset: full · Tools enabled | Moonshot AI ↗ (opens in a new tab)provider reported |
| Humanity's Last ExamStable HLE — text-only subsetAccuracy · Higher is better | 46.9% | question count: 2158 · reported precision decimals: 1 · Kimi K3 (max); original HLE; 2,158 text-only questions; no tools; pass@1; chart labels rounded to one decimal. · source model href: /models/kimi-k3 · subset: text · No tools | Artificial Analysis (opens in a new tab)independent |
| LiveBench: Agentic2.0 / 2026-10-25Accuracy · Higher is better | 26.4% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: agentic | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| LiveBench: Browser Use2.0 / 2026-10-25Accuracy · Higher is better | 75.2% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: browser-use | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| LiveBench: Coding2.0 / 2026-10-25Accuracy · Higher is better | 59.3% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: coding | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| LiveBench: Data Analysis2.0 / 2026-10-25Accuracy · Higher is better | 78.7% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: data-analysis | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| LiveBench: Design2.0 / 2026-10-25Accuracy · Higher is better | 77.2% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: design | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| LiveBench: Reasoning2.0 / 2026-10-25Accuracy · Higher is better | 90.7% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: reasoning | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| LiveBench: Research2.0 / 2026-10-25Accuracy · Higher is better | 79% | release: 2026-10-25 · Highest supported effort unless publisher reports an exception; single run · suite: research | LiveBench ↗ (opens in a new tab)benchmark publisher · 2026-10-25 |
| Terminal-Bench2.1 / Artificial AnalysisPass@1 · Higher is better | 85.019% | Kimi K3 (Max); 89 tasks; Terminus 2; e2b; three repeats | Artificial Analysis (opens in a new tab)independent |
| Terminal-Bench2.1 / Artificial AnalysisPass@1 · Higher is better | 82.397% | Kimi K3 (Low); 89 tasks; Terminus 2; e2b; three repeats | Artificial Analysis (opens in a new tab)independent |
| Terminal-Bench4.0 / mini-swe-agentTask success · Higher is better | 12.626% | harness: mini-swe-agent · repeats: 3 · selection: best reported configuration per release · Kimi K3 (Low) | Artificial Analysis (opens in a new tab)independent |