Claude Opus 5 vs Kimi K3.1 Pro
New ComparisonScores are fetched live from the current database on every visit — nothing is cached.
Winner
Kimi K3.1 Pro
Kimi K3.1 Pro wins this comparison with an overall score of 88 vs 55 for Claude Opus 5, leading 12 of 13 shared benchmarks.
Kimi K3.1 Pro
88
Claude Opus 5
55
Based on 13 of 87 benchmarks scored by all 2 selected models.
Benchmark Scores Comparison
Claude Opus 5
Kimi K3.1 Pro
Claude Opus 5
Score: 40.0
Kimi K3.1 Pro
2.8T params
Coverage matrix
2 models × 87 benchmarks — hover a benchmark name for its category
Shading = score relative to the best score among the selected models:
Low
Mid
High
Missing (—)
The totals row shows how many of the 87 benchmarks each model has scores for.
| Benchmark | Kimi K3.1 Pro Moonshot AI | Claude Opus 5 Anthropic |
|---|---|---|
| PerceptionBench | 68.3 | — |
| GPQA Diamond | 94.2 | 93.7 |
| ArXivMath | — | 71.5 |
| Hy-LifeSearch (Internal) | — | 56.1 |
| Design Bench | 957 | — |
| Automation-Bench | — | 48.7 |
| SWE-bench Pro | — | 79.2 |
| BioMysteryBench | 80.6 | 72.1 |
| BenchCAD w/o tools | 72.8 | — |
| BabyVision w/ python | 95.4 | — |
| Agents' Last Exam | 43.2 | 25.1 |
| Toolathlon Verified | — | 76.5 |
| Hy-Backend 2.0 (Internal) | — | 40.3 |
| Kimi Hallu Bench w/o tools | 5.4 | — |
| Hy-SWE Max Verified (Internal) | — | 70.1 |
| Hy-BrowseComp-Pro2 (Internal) | — | 61.3 |
| Kimi Hallu Bench Hard w/ tools | 8.3 | — |
| ZeroBench_main w/ python (pass@5) | 63 | — |
| OfficeQA Pro | 67.4 | 66.9 |
| GDP.pdf w/ python | 85.7 | — |
| BenchCAD w/ tools | 92.7 | — |
| SWE Atlas - TW | — | 68.7 |
| ChartoGraphy w/ python | 90.2 | — |
| LOCA-Bench (256k) | — | 53.7 |
| SWE Atlas - QnA | — | 57.5 |
| NL2Repo | — | 75.3 |
| MMMU-Pro | 84.1 | — |
| Program Bench | 31.5 | — |
| PostTrainBench | 36.5 | 35 |
| BabyVision | — | 94.1 |
| OSWorld V2.1 | 79.6 | — |
| Apex-Agents | 69.4 | 41.8 |
| WorkSpaceBench | — | 75 |
| Terminal-Bench Science 0.1 | 38.1 | — |
| BankerToolBench | — | 81.9 |
| E-Bench-Code (Internal) | — | 82.7 |
| ExploitGym (6h) | 20.7 | — |
| Frontier Finance | 58.6 | — |
| xdev-bench-hard | 81 | — |
| HLE (with tools) | 70 | 60.9 |
| CharXiv (RQ) | 89 | 83.7 |
| Automation Bench-AA | 73.3 | — |
| GDP.pdf | 34.6 | 21.6 |
| CritPt (no tools) | — | 29.1 |
| BrowseComp | 92.1 | — |
| NL2Repo-Bench | — | 75.3 |
| DRACO | — | 88.6 |
| SWE-bench Multilingual | — | 89.5 |
| Harvey Legal Agent | 23.7 | — |
| OfficeQA Pro v2 | 64.4 | — |
| CharXiv (RQ) w/ python | 93.4 | — |
| Agents' Last Exam (CLI) | 44.2 | — |
| Hy-FinAgentBench (Internal) | — | 82 |
| Humanity's Last Exam | 51.1 | 54.9 |
| HANDBOOK.md | 44.6 | — |
| HLE-Diamond (no tools) | 39.4 | — |
| MMMU-Pro w/ python | 90 | — |
| SUPERChem | — | 76.7 |
| HorizonMath (pass@4) | — | 5.3 |
| Terminal-Bench 2.1 (Terminus-2) | — | 86.7 |
| DeepSWE | — | 68.8 |
| ExploitBench | 74.7 | — |
| BrokenArXiv | — | 77.7 |
| Hy-FinmodelBench v2 (Internal) | — | 66 |
| MathArena Apex 2025 | — | 91.4 |
| Kimi Code Bench 3.0 Long-horizon | 61.2 | — |
| JobBench | — | 68 |
| $OneMillion-Bench | — | 68.1 |
| Kimi Code Bench V2 | 77.1 | — |
| OSWorld 2.0 (Binary) | — | 70.6 |
| DECK-Arena | 1456 | — |
| WideSearch | — | 84 |
| SkillsBench Avg5 | — | 63.7 |
| E-Bench (Internal) | — | 77.8 |
| ProgramBench | — | 39.5 |
| MCP-Atlas | — | 85.7 |
| SWE-Marathon | 51.7 | 50 |
| Terminal-Bench 4.0 | 65.7 | 51.8 |
| SWE Atlas - RF | — | 60 |
| DeepSWE 1.1 | 75.5 | 74 |
| Hy-CompanyBench V2 (Internal) | — | 72.7 |
| Harbor-Index | — | 56.9 |
| Business Operation | 37.8 | — |
| SWE Refactor Bench | — | 31 |
| AA-Briefcase v1.1 | 1678 | — |
| GDPval-AA v2.1 | 1846 | — |
| GDPVal-AA v2 | — | 1831 |
| Covered | 46/87 | 54/87 |
Scorecard
Overall 0–100 score — category strengths (50%), shared-benchmark wins (30%), coverage (20%)
| Model | Overall score | Shared benchmark wins | Coverage | Strongest at | Weakest at |
|---|---|---|---|---|---|
| Kimi K3.1 Pro Winner | 88 | 12 of 13 | 46 of 87 (53.0%) | Long Context | STEM & Reasoning |
| Claude Opus 5 | 55 | 1 of 13 | 54 of 87 (62.0%) | Agentic Coding | Long Context |
Category strengths
Where each model stands, per benchmark category — score relative to the best model, adjusted for benchmark coverage. This also ranks models when they share few benchmarks.
| Category | Kimi K3.1 Pro | Claude Opus 5 |
|---|---|---|
| Agentic Coding 26 benchmarks | 100 | 90.5 |
| General Agents 24 benchmarks | 100 | 66.7 |
| STEM & Reasoning 11 benchmarks | 98.2 | 87.4 |
| Document Understanding 3 benchmarks | 100 | 77 |
| Long Context 2 benchmarks | 100 | 58.1 |
Fully comparable benchmarks
Scored by all 2 selected models (13)
| Benchmark | Category | Claude Opus 5 | Kimi K3.1 Pro |
|---|---|---|---|
| GPQA Diamond | stem_reasoning | 93.7 | 94.2 |
| BioMysteryBench | stem_reasoning | 72.1 | 80.6 |
| Agents' Last Exam | general_agent | 25.1 | 43.2 |
| OfficeQA Pro | general_agent | 66.9 | 67.4 |
| PostTrainBench | coding_agent | 35 | 36.5 |
| Apex-Agents | general_agent | 41.8 | 69.4 |
| HLE (with tools) | stem_reasoning | 60.9 | 70 |
| CharXiv (RQ) | document_understanding | 83.7 | 89 |
| GDP.pdf | long_context | 21.6 | 34.6 |
| Humanity's Last Exam | stem_reasoning | 54.9 | 51.1 |
| SWE-Marathon | coding_agent | 50 | 51.7 |
| Terminal-Bench 4.0 | coding_agent | 51.8 | 65.7 |
| DeepSWE 1.1 | coding_agent | 74 | 75.5 |
Partial coverage benchmarks
Scored by at least one but not all selected models (74) — missing scores show as —
| Benchmark | Category | Claude Opus 5 | Kimi K3.1 Pro |
|---|---|---|---|
| PerceptionBench 1/2 models | vision_language | — | 68.3 |
| ArXivMath 1/2 models | stem_reasoning | 71.5 | — |
| Hy-LifeSearch (Internal) 1/2 models | general_agent | 56.1 | — |
| Design Bench 1/2 models | agentic | — | 957 |
| Automation-Bench 1/2 models | general_agent | 48.7 | — |
| SWE-bench Pro 1/2 models | coding_agent | 79.2 | — |
| BenchCAD w/o tools 1/2 models | coding_agent | — | 72.8 |
| BabyVision w/ python 1/2 models | vision_language | — | 95.4 |
| Toolathlon Verified 1/2 models | general_agent | 76.5 | — |
| Hy-Backend 2.0 (Internal) 1/2 models | coding_agent | 40.3 | — |
| Kimi Hallu Bench w/o tools 1/2 models | safety | — | 5.4 |
| Hy-SWE Max Verified (Internal) 1/2 models | coding_agent | 70.1 | — |
| Hy-BrowseComp-Pro2 (Internal) 1/2 models | general_agent | 61.3 | — |
| Kimi Hallu Bench Hard w/ tools 1/2 models | safety | — | 8.3 |
| ZeroBench_main w/ python (pass@5) 1/2 models | vision_language | — | 63 |
| GDP.pdf w/ python 1/2 models | document_understanding | — | 85.7 |
| BenchCAD w/ tools 1/2 models | coding_agent | — | 92.7 |
| SWE Atlas - TW 1/2 models | coding_agent | 68.7 | — |
| ChartoGraphy w/ python 1/2 models | vision_language | — | 90.2 |
| LOCA-Bench (256k) 1/2 models | long_context | 53.7 | — |
| SWE Atlas - QnA 1/2 models | coding_agent | 57.5 | — |
| NL2Repo 1/2 models | coding_agent | 75.3 | — |
| MMMU-Pro 1/2 models | vision_language | — | 84.1 |
| Program Bench 1/2 models | coding_agent | — | 31.5 |
| BabyVision 1/2 models | vision_language | 94.1 | — |
| OSWorld V2.1 1/2 models | general_agent | — | 79.6 |
| WorkSpaceBench 1/2 models | general_agent | 75 | — |
| Terminal-Bench Science 0.1 1/2 models | coding_agent | — | 38.1 |
| BankerToolBench 1/2 models | general_agent | 81.9 | — |
| E-Bench-Code (Internal) 1/2 models | coding_agent | 82.7 | — |
| ExploitGym (6h) 1/2 models | cybersecurity | — | 20.7 |
| Frontier Finance 1/2 models | domain_finance | — | 58.6 |
| xdev-bench-hard 1/2 models | safety | — | 81 |
| Automation Bench-AA 1/2 models | agentic | — | 73.3 |
| CritPt (no tools) 1/2 models | stem_reasoning | 29.1 | — |
| BrowseComp 1/2 models | general_agent | — | 92.1 |
| NL2Repo-Bench 1/2 models | coding_agent | 75.3 | — |
| DRACO 1/2 models | general_agent | 88.6 | — |
| SWE-bench Multilingual 1/2 models | coding_agent | 89.5 | — |
| Harvey Legal Agent 1/2 models | general_agent | — | 23.7 |
| OfficeQA Pro v2 1/2 models | document_understanding | — | 64.4 |
| CharXiv (RQ) w/ python 1/2 models | vision_language | — | 93.4 |
| Agents' Last Exam (CLI) 1/2 models | general_agent | — | 44.2 |
| Hy-FinAgentBench (Internal) 1/2 models | domain_finance | 82 | — |
| HANDBOOK.md 1/2 models | general_agent | — | 44.6 |
| HLE-Diamond (no tools) 1/2 models | stem_reasoning | — | 39.4 |
| MMMU-Pro w/ python 1/2 models | vision_language | — | 90 |
| SUPERChem 1/2 models | stem_reasoning | 76.7 | — |
| HorizonMath (pass@4) 1/2 models | stem_reasoning | 5.3 | — |
| Terminal-Bench 2.1 (Terminus-2) 1/2 models | coding_agent | 86.7 | — |
| DeepSWE 1/2 models | coding_agent | 68.8 | — |
| ExploitBench 1/2 models | cybersecurity | — | 74.7 |
| BrokenArXiv 1/2 models | stem_reasoning | 77.7 | — |
| Hy-FinmodelBench v2 (Internal) 1/2 models | domain_finance | 66 | — |
| MathArena Apex 2025 1/2 models | stem_reasoning | 91.4 | — |
| Kimi Code Bench 3.0 Long-horizon 1/2 models | coding_agent | — | 61.2 |
| JobBench 1/2 models | general_agent | 68 | — |
| $OneMillion-Bench 1/2 models | general_capabilities | 68.1 | — |
| Kimi Code Bench V2 1/2 models | coding_agent | — | 77.1 |
| OSWorld 2.0 (Binary) 1/2 models | general_agent | 70.6 | — |
| DECK-Arena 1/2 models | agentic | — | 1456 |
| WideSearch 1/2 models | general_agent | 84 | — |
| SkillsBench Avg5 1/2 models | coding_agent | 63.7 | — |
| E-Bench (Internal) 1/2 models | general_agent | 77.8 | — |
| ProgramBench 1/2 models | coding_agent | 39.5 | — |
| MCP-Atlas 1/2 models | general_agent | 85.7 | — |
| SWE Atlas - RF 1/2 models | coding_agent | 60 | — |
| Hy-CompanyBench V2 (Internal) 1/2 models | general_agent | 72.7 | — |
| Harbor-Index 1/2 models | coding_agent | 56.9 | — |
| Business Operation 1/2 models | agentic | — | 37.8 |
| SWE Refactor Bench 1/2 models | coding_agent | 31 | — |
| AA-Briefcase v1.1 1/2 models | general_agent | — | 1678 |
| GDPval-AA v2.1 1/2 models | general_agent | — | 1846 |
| GDPVal-AA v2 1/2 models | general_agent | 1831 | — |
Comparison FAQ
Which model is better, Kimi K3.1 Pro or Claude Opus 5?
Kimi K3.1 Pro wins this comparison with an overall score of 88 vs 55 for Claude Opus 5, leading 12 of 13 shared benchmarks.
Which model is better at Agentic Coding, Kimi K3.1 Pro or Claude Opus 5?
Kimi K3.1 Pro leads the Agentic Coding category (strength score 100/100 across 4 comparable benchmarks).
How reliable is this comparison verdict?
Based on 13 of 87 benchmarks scored by all 2 selected models.