Deepseek v4 Pro High vs GPT-6.1 Sol
Deepseek v4 Pro High 
GPT-6.1 Sol Verdict
- GPT-6.1 Sol (max) leads on quality: 69.3 vs 60.0.
- GPT-6.1 Sol (max) is stronger in coding, human preference, maths, reasoning, agents & tools, composite, knowledge, long context, multimodal.
| Metric | Deepseek v4 Pro High (high) | GPT-6.1 Sol (max) |
|---|---|---|
| BenchLeader Index | 60.0 | 69.3 |
| Coding score | 57.2 | 72.4 |
| Human preference score | 65.7 | 68.1 |
| Maths score | 65.2 | 72.5 |
| Reasoning score | 65.4 | 76.4 |
| Agents & tools score | – | 63.9 |
| Composite score | – | 79.0 |
| Knowledge score | – | 78.8 |
| Long context score | – | 66.8 |
| Multimodal score | – | 66.2 |
| Blended price $/M | – | $4.00 |
| Output speed | – | 56 tok/s |
| Time to first answer | – | 326.9 s |
| Context window | – | 1.1M |
| GPQA Diamond | 90.9% | 95.4% |
| FrontierMath Tiers 1–3 | – | 93.7% |
| FrontierMath Tier 4 | – | 100.0% |
| OTIS Mock AIME | 95.6% | 100.0% |
| SimpleQA Verified | – | 73.9% |
| Terminal-Bench | – | 58.2% |
| SciCode | 46.4% | 54.2% |
| WeirdML | 46.5% | – |
| APEX-Agents | – | 60.0% |
| LMArena Text | 1465 | 1484 |
| LMArena Hard Prompts | 1483 | 1507 |
| LMArena Coding | 1506 | 1545 |
| LMArena WebDev | 1582 | 1755 |
| LMArena Vision | – | 1288 |
| LMArena Agent | – | 11.7 |
| LiveBench | – | 81.6% |
| LiveBench Reasoning | – | 92.6% |
| LiveBench Coding | – | 80.4% |
| LiveBench Agentic Coding | – | 54.5% |
| LiveBench Mathematics | – | 96.8% |
| LiveBench Data Analysis | – | 82.7% |
| LiveBench Language | – | 90.1% |
| LiveBench Instruction Following | – | 74.2% |
| AA Intelligence Index v4.3.2 | – | 51.8 |
| AA-LCR | – | 83.0% |
| MMMU-Pro | – | 86.0% |
| AA-Omniscience | – | 41.5 |
| Humanity's Last Exam (AA) | – | 52.9% |
| SciCode (AA) | – | 54.2% |
| IOI | – | 96.9% |
| Vals Index | – | 61.1 |
| ARC-AGI-1 | – | 96.5% |
| ARC-AGI-2 | – | 94.2% |
| ARC-AGI-3 | – | 96.2% |
| CritPt | – | 31.7% |
| GDPval-AA v2.1 | – | 53.8% |
| Analyst Agent (AA) | – | 50.0% |
| BioMysteryBench | – | 79.6% |
| Code Migration | – | 65.1% |
| CyberBench | – | 39.3% |
| Excel Modeling Benchmark | – | 70.8% |
| Finance Agent v2 | – | 52.0% |
| Harvey's Legal Agent Benchmark | – | 5.4% |
| Legal Research Bench | – | 38.5% |
| MedCode | – | 48.8% |
| MedScribe | – | 86.5% |
| MysteryMechanism | – | 46.4% |
| Public Benefits Bench | – | 59.3% |
| SAGE | – | 46.5% |
| SREBench | – | 50.8% |
| Tax Agent Bench | – | 62.3% |
| Terminal-Bench 4.0 (Vals) | – | 55.0% |
| Vibe Code Bench v1.1 | – | 88.9% |
| LMArena Maths | 1466 | 1488 |
| LMArena Creative Writing | 1446 | 1462 |
| LMArena Instruction Following | 1461 | 1488 |
| LMArena Multi-turn | 1472 | 1487 |
| LMArena Longer Queries | 1475 | 1496 |
| Chess Puzzles | 13.0% | 61.0% |
| EBR-bench | – | 54.3% |
| Mystery Game Puzzles | – | 80.0% |
| CL-bench Life | 13.5% | – |
| Surface Evolver Bench | 40.0% | – |
| ALE-Bench | 1006.1 | – |
| Terminal-Bench 4.0 (AA) | – | 56.1% |
| AutomationBench | – | 64.9% |
| GDP.pdf | – | 31.0% |
| MLCR | – | 33.9% |
| Harvey LAB | – | 6.9% |
| AA-Omniscience: accuracy | – | 62.1% |
| AA-Omniscience: non-hallucination | – | 45.7% |
| AA-Briefcase v1.1 | – | 1557 |
Data as of 2026-10-11. Best configuration of each model; every score links to its source on the model pages.
Deepseek v4 Pro High vs GPT-6.1 Sol: questions
- Is Deepseek v4 Pro High better than GPT-6.1 Sol?
- GPT-6.1 Sol (max) leads on quality: 69.3 vs 60.0. The BenchLeader Index combines every independent quality benchmark; GPT-6.1 Sol (max) is ahead overall as of 2026-10-11, but check the category scores for your use.
- Is Deepseek v4 Pro High better than GPT-6.1 Sol for coding?
- GPT-6.1 Sol scores higher in coding (72 vs 57 on the category index, where 50 is average).