Gemini 3.1 Pro vs Claude Sonnet 4.6 vs DeepSeek V4 Pro vs Claude Opus 5 vs Mistral Large 4 vs SenseNova-Loop-9B-Preview vs Hydrogen-1.0 vs Ripple-65B Preview
New ComparisonScores are fetched live from the current database on every visit — nothing is cached.
Winner
Claude Opus 5
Claude Opus 5 scores higher overall across category strength profiles (74 vs 57 for DeepSeek V4 Pro); only 0 benchmarks are directly comparable.
Claude Opus 5
74.2
DeepSeek V4 Pro
57.3
Gemini 3.1 Pro
57.1
Claude Sonnet 4.6
46.2
Ripple-65B Preview
42.2
Mistral Large 4
37.4
Hydrogen-1.0
0
SenseNova-Loop-9B-Preview
0
Only 0 of 122 benchmarks are scored by every selected model — the verdict leans on category strength profiles and is indicative.
Benchmark Scores Comparison
Gemini 3.1 Pro
Claude Sonnet 4.6
DeepSeek V4 Pro
Claude Opus 5
Mistral Large 4
SenseNova-Loop-9B-Preview
Hydrogen-1.0
Ripple-65B Preview
Gemini 3.1 Pro
Score: 15.0
Claude Sonnet 4.6
Score: 15.0
DeepSeek V4 Pro
Score: 15.0
Claude Opus 5
Score: 40.0
Mistral Large 4
1.0T params
SenseNova-Loop-9B-Preview
10.0B params
Hydrogen-1.0
753.3B params
Ripple-65B Preview
65.0B params
Coverage matrix
8 models × 122 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 122 benchmarks each model has scores for.
| Benchmark | Mistral Large 4 Mistral AI | SenseNova-Loop-9B-Preview sensenova | Hydrogen-1.0 deep-principle | Ripple-65B Preview Ricursive Intelligence | Gemini 3.1 Pro Google | Claude Sonnet 4.6 Anthropic | DeepSeek V4 Pro DeepSeek | Claude Opus 5 Anthropic |
|---|---|---|---|---|---|---|---|---|
| SVG-Bench | — | — | — | — | 59.2 | 64.1 | 61.7 | — |
| Malicious Cyber Request Refusal | 96.6 | — | — | — | — | — | — | — |
| GPQA Diamond | — | — | — | 85.2 | 94.3 | — | 92.8 | 93.7 |
| FrontierCode 2.0 | 19.7 | — | — | — | — | — | — | — |
| ArXivMath | — | — | — | — | — | — | 62.1 | 71.5 |
| LiveCodeBench v6 | — | — | — | — | 91.7 | — | — | — |
| Hy-LifeSearch (Internal) | — | — | — | — | — | — | 46.9 | 56.1 |
| Automation-Bench | — | — | — | — | — | — | 30.4 | 48.7 |
| SWE-bench Pro | — | — | — | 61.42 | 54.2 | — | 55.4 | 79.2 |
| BioMysteryBench | — | — | — | — | — | — | 61.6 | 72.1 |
| Artificial Analysis Coding Agent Index | 48 | — | — | — | — | — | — | — |
| AA-LCR | — | — | — | 75 | — | — | — | — |
| MathVision | — | — | — | — | 89.8 | — | — | — |
| AIME 26 | — | — | — | 95 | 98.2 | — | — | — |
| LiveSQLBench | — | — | — | — | 39.8 | — | — | — |
| AgentWorldBench Terminal | — | — | — | — | 52.47 | 56.98 | — | — |
| Agents' Last Exam | — | — | — | — | — | — | 21.9 | 25.1 |
| Toolathlon Verified | — | — | — | — | 48.8 | — | 74.1 | 76.5 |
| Hy-Backend 2.0 (Internal) | — | — | — | — | — | — | 33.8 | 40.3 |
| OJBench | — | — | — | — | 70.7 | — | — | — |
| Hy-SWE Max Verified (Internal) | — | — | — | — | — | — | 65.7 | 70.1 |
| AgentWorldBench Overall | — | — | — | — | 54.57 | 56.04 | — | — |
| Hy-BrowseComp-Pro2 (Internal) | — | — | — | — | — | — | 46.5 | 61.3 |
| AutomationBench (Public) | — | — | — | 36.4 | — | — | — | — |
| GDPVal | — | — | — | — | 57.8 | 75.7 | 70.3 | — |
| OfficeQA Pro | — | — | — | — | 18.1 | — | 65.4 | 66.9 |
| Apex-Shortlist (no tools) | — | — | — | — | 89.1 | — | — | — |
| DeepSearch QA | — | — | — | — | 81.9 | — | — | — |
| StrongREJECT v2 (adaptive) | — | — | — | 97.5 | — | — | — | — |
| SWE Atlas - TW | — | — | — | — | 29.8 | 31.8 | 45.6 | 68.7 |
| CL-bench | — | — | — | — | 21.1 | — | — | — |
| LOCA-Bench (256k) | — | — | — | — | — | — | — | 53.7 |
| KernelBench Hard | — | — | — | — | 18.6 | — | — | — |
| SWE-bench Pro v2.0 Hard | — | — | — | 79.1 | — | — | — | — |
| AgentWorldBench Web | — | — | — | — | 52.83 | 50.78 | — | — |
| SWE Atlas - QnA | — | — | — | — | 13.5 | 31.2 | 53.4 | 57.5 |
| IMO 2025 | — | — | — | — | 42.4 | — | — | — |
| NL2Repo | — | — | — | — | 33.4 | — | 61.5 | 75.3 |
| MMMU-Pro | — | — | — | — | 80.5 | 74.5 | — | — |
| PostTrainBench | — | — | — | — | 15.2 | — | 24.5 | 35 |
| BabyVision | — | — | — | — | 51.6 | — | — | 94.1 |
| Claw-Eval Avg | — | — | — | — | 57.8 | 68.3 | 58.4 | — |
| VideoMMMU | — | — | — | — | 87.9 | — | — | — |
| HMMT Nov 25 | — | — | — | — | 94.8 | — | — | — |
| Apex-Agents | — | — | — | — | 33.4 | 26.2 | 32.4 | 41.8 |
| Artificial Analysis Intelligence Index v5 | 26 | — | — | — | — | — | — | — |
| WorkSpaceBench | — | — | — | — | — | — | 65.4 | 75 |
| MMLU-Pro | — | — | — | 84.2 | 91 | — | — | — |
| CodeForces | — | — | — | — | 3052 | — | — | — |
| Terminal-Bench Science 0.1 | 4.3 | — | — | — | — | — | — | — |
| BankerToolBench | — | — | — | — | 67 | — | 73.1 | 81.9 |
| E-Bench-Code (Internal) | — | — | — | — | — | — | 64.3 | 82.7 |
| OSWorld-Verified | — | — | — | — | 76.2 | 72.5 | — | — |
| AgentWorldBench Search | — | — | — | — | 30.21 | 28.79 | — | — |
| HLE (with tools) | — | — | — | — | 51.4 | — | 60 | 60.9 |
| CharXiv (RQ) | — | — | — | — | 80.2 | — | — | 83.7 |
| FrontierSWE | — | — | — | — | 39.6 | — | — | — |
| GDP.pdf | — | — | — | — | — | — | — | 21.6 |
| AIME 2024 | — | — | — | 97.1 | — | — | — | — |
| CritPt (no tools) | — | — | — | — | 17.7 | — | 18 | 29.1 |
| BrowseComp | 84.5 | — | — | — | 85.9 | 74.7 | 83.4 | — |
| NL2Repo-Bench | — | — | — | — | 21.6 | — | 35.5 | 75.3 |
| DRACO | — | — | — | — | — | 75.8 | 77.3 | 88.6 |
| MCPMark | — | — | — | — | 55.9 | — | — | — |
| SWE-bench Multilingual | — | — | — | — | 76.9 | — | 77.3 | 89.5 |
| MMLU | — | — | — | 89.6 | — | — | — | — |
| PaperBench | — | — | — | — | 46.7 | — | — | — |
| YC-Bench | — | — | — | — | 1100000 | 100000 | 500000 | — |
| Hy-FinAgentBench (Internal) | — | — | — | — | — | — | 78.5 | 82 |
| IFBench | — | — | — | 79.2 | — | — | — | — |
| SWE-bench Verified | — | — | — | 76.5 | 80.6 | 79.6 | 80.6 | — |
| IMOAnswerBench | — | — | — | — | 81 | — | — | — |
| Humanity's Last Exam | — | — | — | — | 45 | — | 42.7 | 54.9 |
| Terminal-Bench 2.0 | — | — | — | — | 68.5 | — | — | — |
| Artificial Analysis Cyber Index | 51 | — | — | — | — | — | — | — |
| SpreadSheetBench-v1 | — | — | — | — | 56.1 | — | 84.9 | — |
| SUPERChem | — | — | — | — | — | — | 62 | 76.7 |
| VIBE-V2 | — | — | — | — | 28 | 42.8 | 45 | — |
| HorizonMath (pass@4) | — | — | — | — | — | — | 4.42 | 5.3 |
| Terminal-Bench 2.1 (Terminus-2) | — | — | — | 80.5 | 74 | — | 87.9 | 86.7 |
| DeepSWE | — | — | — | — | — | — | 62.7 | 68.8 |
| BrokenArXiv | — | — | — | — | — | — | 43.1 | 77.7 |
| OmniDocBench 1.5 | — | — | — | — | 88.1 | 86.9 | — | — |
| B3 Agent Security Benchmark (Attack Resistance) | 97.1 | — | — | — | — | — | — | — |
| HMMT Feb 26 | — | — | — | 90.5 | 87.3 | — | — | — |
| AgentWorldBench OS | — | — | — | — | 66.92 | 63.17 | — | — |
| AgentWorldBench MCP | — | — | — | — | 59.07 | 70 | — | — |
| Hy-FinmodelBench v2 (Internal) | — | — | — | — | — | — | 51.3 | 66 |
| MathArena Apex 2025 | — | — | — | — | — | — | 66.3 | 91.4 |
| Terminal Bench 2.1 | — | — | — | 80.5 | 70.3 | — | 59.6 | — |
| AIME 2025 | — | — | — | 98.3 | — | — | — | — |
| AgentWorldBench SWE | — | — | — | — | 59.07 | 64.52 | — | — |
| JobBench | — | — | — | — | — | — | 54.1 | 68 |
| $OneMillion-Bench | — | — | — | — | — | — | 62 | 68.1 |
| SciCode (subtask) | — | — | — | — | 58.9 | — | — | — |
| USAMO 2026 | — | — | — | — | 74.4 | — | — | — |
| Tool Decathlon | — | — | — | — | 48.8 | — | — | — |
| OSWorld 2.0 (Binary) | — | — | — | — | — | — | — | 70.6 |
| WideSearch | — | — | — | — | — | — | 81.8 | 84 |
| VideoMME (w sub.) | — | — | — | — | 87.9 | — | — | — |
| SkillsBench Avg5 | — | — | — | — | — | — | 65 | 63.7 |
| TauBench V3 Banking | — | — | — | 31.27 | — | — | — | — |
| Claw-Eval Pass^3 | — | — | — | — | 57.8 | — | — | — |
| E-Bench (Internal) | — | — | — | — | — | — | 61.3 | 77.8 |
| ProgramBench | — | — | — | — | 39.5 | — | 15.5 | 39.5 |
| MCP-Atlas | — | — | — | 71.67 | 69.2 | 61.3 | 73.6 | 85.7 |
| SWE-Marathon | — | — | — | — | 4 | — | 19 | 50 |
| Terminal-Bench 2.1 (Best Reported Harness) | — | — | — | — | 70.7 | — | — | — |
| Cybergym | — | — | — | — | — | — | 83.3 | — |
| Terminal-Bench 4.0 | 41 | — | — | — | — | — | — | 51.8 |
| SWE Atlas - RF | — | — | — | — | — | — | 48.6 | 60 |
| V-Star | — | — | — | — | 96.9 | — | — | — |
| AgentWorldBench Android | — | — | — | — | 61.4 | 58.03 | — | — |
| DeepSWE 1.1 | — | — | — | 44.3 | 10 | — | — | 74 |
| FORTRESS Benign | — | — | — | 95.5 | — | — | — | — |
| Hy-CompanyBench V2 (Internal) | — | — | — | — | — | — | 64.4 | 72.7 |
| Harbor-Index | — | — | — | — | — | — | 36.9 | 56.9 |
| SWE-efficiency | — | — | — | — | 19.7 | — | — | — |
| SWE Refactor Bench | — | — | — | — | — | — | — | 31 |
| SciCode | — | — | — | 39.4 | — | — | — | — |
| FinWorkBench (Finch) | 73.3 | — | — | — | — | — | — | — |
| GDPVal-AA v2 | — | — | — | — | 1314 | — | 1580 | 1831 |
| Covered | 10/122 | 0/122 | 0/122 | 21/122 | 71/122 | 23/122 | 56/122 | 54/122 |
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 |
|---|---|---|---|---|---|
| Claude Opus 5 Winner | 74.2 | — | 54 of 122 (44.0%) | General Capabilities | Document Understanding |
| DeepSeek V4 Pro | 57.3 | — | 56 of 122 (46.0%) | General Capabilities | STEM & Reasoning |
| Gemini 3.1 Pro | 57.1 | — | 71 of 122 (58.0%) | Knowledge | Agentic Coding |
| Claude Sonnet 4.6 | 46.2 | — | 23 of 122 (19.0%) | Document Understanding | Agentic Coding |
| Ripple-65B Preview | 42.2 | — | 21 of 122 (17.0%) | Knowledge | General Agents |
| Mistral Large 4 | 37.4 | — | 10 of 122 (8.0%) | General Agents | Agentic Coding |
| Hydrogen-1.0 | 0 | — | 0 of 122 (0.0%) | — | — |
| SenseNova-Loop-9B-Preview | 0 | — | 0 of 122 (0.0%) | — | — |
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 | Claude Opus 5 | DeepSeek V4 Pro | Gemini 3.1 Pro | Claude Sonnet 4.6 | Ripple-65B Preview | Mistral Large 4 | Hydrogen-1.0 | SenseNova-Loop-9B-Preview |
|---|---|---|---|---|---|---|---|---|
| General Agents 38 benchmarks | 79.8 | 59.2 | 60 | 59 | 45.7 | 55.6 | — | — |
| Agentic Coding 37 benchmarks | 90.4 | 67.3 | 41.8 | 52.3 | 54.2 | 41.2 | — | — |
| STEM & Reasoning 23 benchmarks | 92.4 | 55.4 | 66.7 | — | 61.5 | — | — | — |
| Vision & Language 4 benchmarks | 77.5 | — | 73.9 | 68.9 | — | — | — | — |
| Document Understanding 2 benchmarks | 77.5 | — | 90.1 | 76.4 | — | — | — | — |
| Finance 2 benchmarks | 100 | 66.3 | — | — | — | — | — | — |
| Knowledge 2 benchmarks | — | — | 100 | — | 78.1 | — | — | — |
| General Capabilities 1 benchmark | 100 | 76.8 | — | — | — | — | — | — |
Partial coverage benchmarks
Scored by at least one but not all selected models (122) — missing scores show as —
| Benchmark | Category | Gemini 3.1 Pro | Claude Sonnet 4.6 | DeepSeek V4 Pro | Claude Opus 5 | Mistral Large 4 | SenseNova-Loop-9B-Preview | Hydrogen-1.0 | Ripple-65B Preview |
|---|---|---|---|---|---|---|---|---|---|
| SVG-Bench 3/8 models | coding_agent | 59.2 | 64.1 | 61.7 | — | — | — | — | — |
| Malicious Cyber Request Refusal 1/8 models | composite | — | — | — | — | 96.6 | — | — | — |
| GPQA Diamond 4/8 models | stem_reasoning | 94.3 | — | 92.8 | 93.7 | — | — | — | 85.2 |
| FrontierCode 2.0 1/8 models | coding_agent | — | — | — | — | 19.7 | — | — | — |
| ArXivMath 2/8 models | stem_reasoning | — | — | 62.1 | 71.5 | — | — | — | — |
| LiveCodeBench v6 1/8 models | stem_reasoning | 91.7 | — | — | — | — | — | — | — |
| Hy-LifeSearch (Internal) 2/8 models | general_agent | — | — | 46.9 | 56.1 | — | — | — | — |
| Automation-Bench 2/8 models | general_agent | — | — | 30.4 | 48.7 | — | — | — | — |
| SWE-bench Pro 4/8 models | coding_agent | 54.2 | — | 55.4 | 79.2 | — | — | — | 61.42 |
| BioMysteryBench 2/8 models | stem_reasoning | — | — | 61.6 | 72.1 | — | — | — | — |
| Artificial Analysis Coding Agent Index 1/8 models | composite | — | — | — | — | 48 | — | — | — |
| AA-LCR 1/8 models | long_context | — | — | — | — | — | — | — | 75 |
| MathVision 1/8 models | vision_language | 89.8 | — | — | — | — | — | — | — |
| AIME 26 2/8 models | stem_reasoning | 98.2 | — | — | — | — | — | — | 95 |
| LiveSQLBench 1/8 models | coding_agent | 39.8 | — | — | — | — | — | — | — |
| AgentWorldBench Terminal 2/8 models | general_agent | 52.47 | 56.98 | — | — | — | — | — | — |
| Agents' Last Exam 2/8 models | general_agent | — | — | 21.9 | 25.1 | — | — | — | — |
| Toolathlon Verified 3/8 models | general_agent | 48.8 | — | 74.1 | 76.5 | — | — | — | — |
| Hy-Backend 2.0 (Internal) 2/8 models | coding_agent | — | — | 33.8 | 40.3 | — | — | — | — |
| OJBench 1/8 models | stem_reasoning | 70.7 | — | — | — | — | — | — | — |
| Hy-SWE Max Verified (Internal) 2/8 models | coding_agent | — | — | 65.7 | 70.1 | — | — | — | — |
| AgentWorldBench Overall 2/8 models | general_agent | 54.57 | 56.04 | — | — | — | — | — | — |
| Hy-BrowseComp-Pro2 (Internal) 2/8 models | general_agent | — | — | 46.5 | 61.3 | — | — | — | — |
| AutomationBench (Public) 1/8 models | general_agent | — | — | — | — | — | — | — | 36.4 |
| GDPVal 3/8 models | general_agent | 57.8 | 75.7 | 70.3 | — | — | — | — | — |
| OfficeQA Pro 3/8 models | general_agent | 18.1 | — | 65.4 | 66.9 | — | — | — | — |
| Apex-Shortlist (no tools) 1/8 models | stem_reasoning | 89.1 | — | — | — | — | — | — | — |
| DeepSearch QA 1/8 models | general_agent | 81.9 | — | — | — | — | — | — | — |
| StrongREJECT v2 (adaptive) 1/8 models | safety | — | — | — | — | — | — | — | 97.5 |
| SWE Atlas - TW 4/8 models | coding_agent | 29.8 | 31.8 | 45.6 | 68.7 | — | — | — | — |
| CL-bench 1/8 models | coding_agent | 21.1 | — | — | — | — | — | — | — |
| LOCA-Bench (256k) 1/8 models | long_context | — | — | — | 53.7 | — | — | — | — |
| KernelBench Hard 1/8 models | coding_agent | 18.6 | — | — | — | — | — | — | — |
| SWE-bench Pro v2.0 Hard 1/8 models | coding_agent | — | — | — | — | — | — | — | 79.1 |
| AgentWorldBench Web 2/8 models | general_agent | 52.83 | 50.78 | — | — | — | — | — | — |
| SWE Atlas - QnA 4/8 models | coding_agent | 13.5 | 31.2 | 53.4 | 57.5 | — | — | — | — |
| IMO 2025 1/8 models | stem_reasoning | 42.4 | — | — | — | — | — | — | — |
| NL2Repo 3/8 models | coding_agent | 33.4 | — | 61.5 | 75.3 | — | — | — | — |
| MMMU-Pro 2/8 models | vision_language | 80.5 | 74.5 | — | — | — | — | — | — |
| PostTrainBench 3/8 models | coding_agent | 15.2 | — | 24.5 | 35 | — | — | — | — |
| BabyVision 2/8 models | vision_language | 51.6 | — | — | 94.1 | — | — | — | — |
| Claw-Eval Avg 3/8 models | coding_agent | 57.8 | 68.3 | 58.4 | — | — | — | — | — |
| VideoMMMU 1/8 models | video_understanding | 87.9 | — | — | — | — | — | — | — |
| HMMT Nov 25 1/8 models | stem_reasoning | 94.8 | — | — | — | — | — | — | — |
| Apex-Agents 4/8 models | general_agent | 33.4 | 26.2 | 32.4 | 41.8 | — | — | — | — |
| Artificial Analysis Intelligence Index v5 1/8 models | composite | — | — | — | — | 26 | — | — | — |
| WorkSpaceBench 2/8 models | general_agent | — | — | 65.4 | 75 | — | — | — | — |
| MMLU-Pro 2/8 models | knowledge | 91 | — | — | — | — | — | — | 84.2 |
| CodeForces 1/8 models | stem_reasoning | 3052 | — | — | — | — | — | — | — |
| Terminal-Bench Science 0.1 1/8 models | coding_agent | — | — | — | — | 4.3 | — | — | — |
| BankerToolBench 3/8 models | general_agent | 67 | — | 73.1 | 81.9 | — | — | — | — |
| E-Bench-Code (Internal) 2/8 models | coding_agent | — | — | 64.3 | 82.7 | — | — | — | — |
| OSWorld-Verified 2/8 models | general_agent | 76.2 | 72.5 | — | — | — | — | — | — |
| AgentWorldBench Search 2/8 models | general_agent | 30.21 | 28.79 | — | — | — | — | — | — |
| HLE (with tools) 3/8 models | stem_reasoning | 51.4 | — | 60 | 60.9 | — | — | — | — |
| CharXiv (RQ) 2/8 models | document_understanding | 80.2 | — | — | 83.7 | — | — | — | — |
| FrontierSWE 1/8 models | coding_agent | 39.6 | — | — | — | — | — | — | — |
| GDP.pdf 1/8 models | long_context | — | — | — | 21.6 | — | — | — | — |
| AIME 2024 1/8 models | stem_reasoning | — | — | — | — | — | — | — | 97.1 |
| CritPt (no tools) 3/8 models | stem_reasoning | 17.7 | — | 18 | 29.1 | — | — | — | — |
| BrowseComp 4/8 models | general_agent | 85.9 | 74.7 | 83.4 | — | 84.5 | — | — | — |
| NL2Repo-Bench 3/8 models | coding_agent | 21.6 | — | 35.5 | 75.3 | — | — | — | — |
| DRACO 3/8 models | general_agent | — | 75.8 | 77.3 | 88.6 | — | — | — | — |
| MCPMark 1/8 models | general_agent | 55.9 | — | — | — | — | — | — | — |
| SWE-bench Multilingual 3/8 models | coding_agent | 76.9 | — | 77.3 | 89.5 | — | — | — | — |
| MMLU 1/8 models | knowledge | — | — | — | — | — | — | — | 89.6 |
| PaperBench 1/8 models | coding_agent | 46.7 | — | — | — | — | — | — | — |
| YC-Bench 3/8 models | general_agent | 1100000 | 100000 | 500000 | — | — | — | — | — |
| Hy-FinAgentBench (Internal) 2/8 models | domain_finance | — | — | 78.5 | 82 | — | — | — | — |
| IFBench 1/8 models | instruction_following | — | — | — | — | — | — | — | 79.2 |
| SWE-bench Verified 4/8 models | coding_agent | 80.6 | 79.6 | 80.6 | — | — | — | — | 76.5 |
| IMOAnswerBench 1/8 models | stem_reasoning | 81 | — | — | — | — | — | — | — |
| Humanity's Last Exam 3/8 models | stem_reasoning | 45 | — | 42.7 | 54.9 | — | — | — | — |
| Terminal-Bench 2.0 1/8 models | coding_agent | 68.5 | — | — | — | — | — | — | — |
| Artificial Analysis Cyber Index 1/8 models | composite | — | — | — | — | 51 | — | — | — |
| SpreadSheetBench-v1 2/8 models | general_agent | 56.1 | — | 84.9 | — | — | — | — | — |
| SUPERChem 2/8 models | stem_reasoning | — | — | 62 | 76.7 | — | — | — | — |
| VIBE-V2 3/8 models | coding_agent | 28 | 42.8 | 45 | — | — | — | — | — |
| HorizonMath (pass@4) 2/8 models | stem_reasoning | — | — | 4.42 | 5.3 | — | — | — | — |
| Terminal-Bench 2.1 (Terminus-2) 4/8 models | coding_agent | 74 | — | 87.9 | 86.7 | — | — | — | 80.5 |
| DeepSWE 2/8 models | coding_agent | — | — | 62.7 | 68.8 | — | — | — | — |
| BrokenArXiv 2/8 models | stem_reasoning | — | — | 43.1 | 77.7 | — | — | — | — |
| OmniDocBench 1.5 2/8 models | document_understanding | 88.1 | 86.9 | — | — | — | — | — | — |
| B3 Agent Security Benchmark (Attack Resistance) 1/8 models | general_agent | — | — | — | — | 97.1 | — | — | — |
| HMMT Feb 26 2/8 models | stem_reasoning | 87.3 | — | — | — | — | — | — | 90.5 |
| AgentWorldBench OS 2/8 models | general_agent | 66.92 | 63.17 | — | — | — | — | — | — |
| AgentWorldBench MCP 2/8 models | general_agent | 59.07 | 70 | — | — | — | — | — | — |
| Hy-FinmodelBench v2 (Internal) 2/8 models | domain_finance | — | — | 51.3 | 66 | — | — | — | — |
| MathArena Apex 2025 2/8 models | stem_reasoning | — | — | 66.3 | 91.4 | — | — | — | — |
| Terminal Bench 2.1 3/8 models | coding_agent | 70.3 | — | 59.6 | — | — | — | — | 80.5 |
| AIME 2025 1/8 models | stem_reasoning | — | — | — | — | — | — | — | 98.3 |
| AgentWorldBench SWE 2/8 models | general_agent | 59.07 | 64.52 | — | — | — | — | — | — |
| JobBench 2/8 models | general_agent | — | — | 54.1 | 68 | — | — | — | — |
| $OneMillion-Bench 2/8 models | general_capabilities | — | — | 62 | 68.1 | — | — | — | — |
| SciCode (subtask) 1/8 models | stem_reasoning | 58.9 | — | — | — | — | — | — | — |
| USAMO 2026 1/8 models | stem_reasoning | 74.4 | — | — | — | — | — | — | — |
| Tool Decathlon 1/8 models | general_agent | 48.8 | — | — | — | — | — | — | — |
| OSWorld 2.0 (Binary) 1/8 models | general_agent | — | — | — | 70.6 | — | — | — | — |
| WideSearch 2/8 models | general_agent | — | — | 81.8 | 84 | — | — | — | — |
| VideoMME (w sub.) 1/8 models | video_understanding | 87.9 | — | — | — | — | — | — | — |
| SkillsBench Avg5 2/8 models | coding_agent | — | — | 65 | 63.7 | — | — | — | — |
| TauBench V3 Banking 1/8 models | general_agent | — | — | — | — | — | — | — | 31.27 |
| Claw-Eval Pass^3 1/8 models | coding_agent | 57.8 | — | — | — | — | — | — | — |
| E-Bench (Internal) 2/8 models | general_agent | — | — | 61.3 | 77.8 | — | — | — | — |
| ProgramBench 3/8 models | coding_agent | 39.5 | — | 15.5 | 39.5 | — | — | — | — |
| MCP-Atlas 5/8 models | general_agent | 69.2 | 61.3 | 73.6 | 85.7 | — | — | — | 71.67 |
| SWE-Marathon 3/8 models | coding_agent | 4 | — | 19 | 50 | — | — | — | — |
| Terminal-Bench 2.1 (Best Reported Harness) 1/8 models | coding_agent | 70.7 | — | — | — | — | — | — | — |
| Cybergym 1/8 models | general_agent | — | — | 83.3 | — | — | — | — | — |
| Terminal-Bench 4.0 2/8 models | coding_agent | — | — | — | 51.8 | 41 | — | — | — |
| SWE Atlas - RF 2/8 models | coding_agent | — | — | 48.6 | 60 | — | — | — | — |
| V-Star 1/8 models | vision_language | 96.9 | — | — | — | — | — | — | — |
| AgentWorldBench Android 2/8 models | general_agent | 61.4 | 58.03 | — | — | — | — | — | — |
| DeepSWE 1.1 3/8 models | coding_agent | 10 | — | — | 74 | — | — | — | 44.3 |
| FORTRESS Benign 1/8 models | safety | — | — | — | — | — | — | — | 95.5 |
| Hy-CompanyBench V2 (Internal) 2/8 models | general_agent | — | — | 64.4 | 72.7 | — | — | — | — |
| Harbor-Index 2/8 models | coding_agent | — | — | 36.9 | 56.9 | — | — | — | — |
| SWE-efficiency 1/8 models | coding_agent | 19.7 | — | — | — | — | — | — | — |
| SWE Refactor Bench 1/8 models | coding_agent | — | — | — | 31 | — | — | — | — |
| SciCode 1/8 models | — | — | — | — | — | — | — | — | 39.4 |
| FinWorkBench (Finch) 1/8 models | general_agent | — | — | — | — | 73.3 | — | — | — |
| GDPVal-AA v2 3/8 models | general_agent | 1314 | — | 1580 | 1831 | — | — | — | — |
Comparison FAQ
Which model scores highest in this comparison?
Claude Opus 5 scores higher overall across category strength profiles (74 vs 57 for DeepSeek V4 Pro); only 0 benchmarks are directly comparable.
Which model is strongest at General Agents?
Claude Opus 5 leads the General Agents category (strength score 79.8/100 across 29 comparable benchmarks).
How reliable is this comparison verdict?
Only 0 of 122 benchmarks are scored by every selected model — the verdict leans on category strength profiles and is indicative.