Kimi K3
| Released | 2026-07-16 |
|---|---|
| Price per 1M | $3.00 in · $15.00 out |
| Overall score | 60.5 · rank 23 |
| Context window | 1,048,576 tokens |
| Provider | Moonshot AI |
| Weights | Open weights |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 60.5 | 23 | 56 |
| Overall: Coding | 59.7 | 31 | 41 |
| Overall: Reasoning | 60.6 | 30 | 19 |
| Overall: Agentic work | 59.8 | 34 | 38 |
| Overall: Frontend | 59.5 | 26 | 17 |
| Coding | 59.4 | 34 | 13 |
| Agents | 59.4 | 37 | 14 |
| Reasoning | 62.1 | 26 | 11 |
| Math | 60.1 | 29 | 5 |
| Knowledge | 58.3 | 41 | 3 |
| Frontend | 59.3 | 27 | 3 |
| General | 65.6 | 6 | 6 |
| Multimodal | 61.1 | 12 | 1 |
| Value vs price | +7.4 | 48 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 61.8 ±1.5 | 22 |
| High | 58.2 ±3.6 | 6 |
| Low | 54.4 ±4.0 | 8 |
| Max | 59.9 ±0.9 | 34 |
| Non-reasoning | 60.5 ±5.1 | 1 |
| Thinking | 61.4 ±7.1 | 1 |
Evidence
Coding 59.4 ±1.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierSWE | 25.9% | Epoch AI | 2026-11-11 |
| DeepSWE | 68.5% | Epoch AI | 2026-11-11 |
| FrontierCode | 44.2% | Epoch AI | 2026-11-11 |
| Code Migration v1.1 | 28.6% | Vals AI | 2026-11-09 |
| LiveBench 2.0 Coding | 53.1% | LiveBench | 2026-11-06 |
| WeirdML v3 | 7.5% | Epoch AI | 2026-11-11 |
| MirrorCode | 54.4% | Epoch AI | 2026-09-03 |
| IOI | 48.9% | Vals AI | 2026-10-29 |
| SWE-bench Multimodal | 44.2% | Vals AI | 2026-11-09 |
| SWE-bench Verified (Vals run) | 93.4% | Vals AI | 2026-09-01 |
| SciCode | 59.5% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 87.2% | Vals AI | 2026-09-01 |
| ProgramBench | 2.0% | Vals AI | 2026-11-09 |
Agents 59.4 ±1.2
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Terminal-Bench 4.0 (Vals run) | 17.2% | Vals AI | 2026-11-09 |
| Terminal-Bench-Science | 1.4% | Vals AI | 2026-11-09 |
| APEX-Agents | 47.8% | Epoch AI | 2026-11-11 |
| LiveBench 2.0 Agentic | 36.0% | LiveBench | 2026-11-06 |
| LiveBench 2.0 Browser Use | 75.2% | LiveBench | 2026-11-06 |
| Terminal-Bench 2.1 (Vals run) | 80.9% | Vals AI | 2026-10-01 |
| tau2-bench Banking Knowledge | 37.1% | tau2-bench | 2026-07-24 |
| EMB v1.1 (Excel financial modeling) | 71.3% | Vals AI | 2026-11-09 |
| LiveBench 2.0 Research | 79.8% | LiveBench | 2026-11-06 |
| Tax Agent Bench | 33.4% | Vals AI | 2026-11-09 |
| Finance Agent v2.1 | 60.0% | Vals AI | 2026-11-09 |
| Vending-Bench 2 | $5,165 | Epoch AI | 2026-11-11 |
| HLAB-Vals 1.1 (legal agent) | 12.9% | Vals AI | 2026-11-09 |
| CyberBench v1.1 | 75.2% | Vals AI | 2026-11-05 |
Reasoning 62.1 ±1.5
| Benchmark | Result | Source | Date |
|---|---|---|---|
| ARC-AGI-2 | 60.4% | Epoch AI | 2026-11-11 |
| ARC-AGI-2 (ARC Prize verified) | 60.4% | ARC Prize | 2026-11-11 |
| Extended NYT Connections | 93.6% | Lech Mazur | 2026-11-10 |
| CritPt | 23.4% | Epoch AI | 2026-11-11 |
| ARC-AGI-1 (ARC Prize verified) | 94.5% | ARC Prize | 2026-11-11 |
| ARC-AGI-1 | 94.5% | Epoch AI | 2026-11-11 |
| GPQA Diamond | 93.1% | Epoch AI | 2026-07-16 |
| GPQA Diamond (Vals run) | 92.9% | Vals AI | 2026-09-01 |
| BioMysteryBench | 72.6% | Vals AI | 2026-11-09 |
| Chess Puzzles | 37.0% | Epoch AI | 2026-10-13 |
| SimpleBench | 60.7% | SimpleBench | 2026-07-17 |
| SimpleBench | 60.7% | Epoch AI | 2026-11-11 |
| LiveBench 2.0 Reasoning | 90.7% | LiveBench | 2026-11-06 |
| Mystery Game Puzzles | 26.0% | Epoch AI | 2026-07-29 |
| LiveBench 2.0 Data Analysis | 78.7% | LiveBench | 2026-11-06 |
Math 60.1 ±1.1
| Benchmark | Result | Source | Date |
|---|---|---|---|
| FrontierMath Tiers 1-3 | 72.2% | Epoch AI | 2026-07-17 |
| FrontierMath Tier 4 | 39.0% | Epoch AI | 2026-07-17 |
| OTIS Mock AIME 2024-2025 | 97.2% | Epoch AI | 2026-07-16 |
| ProofBench v1.1 | 87.0% | Vals AI | 2026-11-09 |
| MathArena Final-Answer Competitions | 87.8% | MathArena | 2026-11-11 |
| ProofBench | 87.0% | Epoch AI | 2026-11-11 |
Knowledge 58.3 ±3.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| SimpleQA Verified | 50.6% | Epoch AI | 2026-08-27 |
| MMLU-Pro (Vals run) | 88.0% | Vals AI | 2026-09-01 |
| Legal Research Bench | 46.2% | Vals AI | 2026-11-09 |
Frontend 59.3 ±2.2
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LiveBench 2.0 Design | 72.5% | LiveBench | 2026-11-06 |
| Vibe Code Bench v1.1 | 85.0% | Vals AI | 2026-11-09 |
| LMArena WebDev | 1,657 | LMArena | 2026-11-09 |
General 65.6 ±2.1
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LLM Debate Benchmark | 1,691 | Lech Mazur | 2026-10-26 |
| EQ-Bench 4 | 1,319 | EQ-Bench | 2026-10-24 |
| EQ-Bench Creative Writing v3 | 2,083 | EQ-Bench | 2026-11-11 |
| LMArena Text | 1,487 | LMArena | 2026-11-09 |
| Creative Story-Writing (Lech Mazur) | 81.0% | Lech Mazur | 2026-10-23 |
| EQ-Bench Longform Creative Writing | 79.6% | EQ-Bench | 2026-11-11 |
Multimodal 61.1 ±2.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMMU-Pro (Vals run) | 88.2% | Vals AI | 2026-09-01 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| ARC-AGI-1 | 65.7% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-1 | 86.7% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-1 (ARC Prize verified) | 65.7% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-1 (ARC Prize verified) | 86.7% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-2 | 12.4% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-2 | 55.0% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-2 (ARC Prize verified) | 12.4% | ARC Prize | A stronger configuration result was used |
| ARC-AGI-2 (ARC Prize verified) | 55.0% | ARC Prize | A stronger configuration result was used |
| Chess Puzzles | 18.0% | Epoch AI | A stronger configuration result was used |
| Chess Puzzles | 36.0% | Epoch AI | A stronger configuration result was used |
| CritPt | 3.1% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 84.9% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 91.9% | Epoch AI | A stronger configuration result was used |
| Mystery Game Puzzles | 12.0% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 68.9% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 93.3% | Epoch AI | A stronger configuration result was used |
| SciCode | 58.7% | Epoch AI | A stronger configuration result was used |
| SciCode | 52.7% | Epoch AI | A stronger configuration result was used |
Names matched to this model
kimi-k3