Model comparison
DeepSeek V4 Pro vs GPT-6 Sol
GPT-6 Sol is the stronger model overall, scoring 61.8 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 4.0× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Last verified . 39 shared benchmarks.
DeepSeekSummary
- They share 39 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 3 categories and GPT-6 Sol in 6 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Sol leads 87.2 to 64.8.
- The biggest single-benchmark swing is FrontierMath Tier 4: 26.8% for DeepSeek V4 Pro and 90% for GPT-6 Sol.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $2 / $10 for GPT-6 Sol.
- GPT-6 Sol accepts more context: 1.05M tokens versus 1M.
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Pro | GPT-6 Sol | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 54.3 | 61.8 |
| Released | 2026-04-24 | 2026-09-22 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 393K | 128K |
| Input $ / M tokens | $0.66 | $2 |
| Output $ / M tokens | $1.98 | $10 |
| Results tracked | 48 | 45 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Sol leads
DeepSeek V4 Pro: 52.4 (#34), GPT-6 Sol: 60.1 (#11)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| FrontierCode | 28.6% | 49.3% |
| LMArena WebDev | 1582 | 1688 |
| SciCode | 51% | 57.6% |
| LMArena Coding | 1470 | 1447 |
| ALE-Bench | 1,403 | 2,462 |
| SWE-bench Verified | 77.6% | — |
| DeepSWE | — | 68.8% |
| WeirdML | 66.2% | — |
Agentic & Tool Use GPT-6 Sol leads
DeepSeek V4 Pro: 32.8 (#58), GPT-6 Sol: 37.2 (#36)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| APEX-Agents | 47.3% | 54.3% |
| Vending-Bench 2 | 3,285 | 14,428 |
| GDP.pdf | — | 26.4% |
Reasoning GPT-6 Sol leads
DeepSeek V4 Pro: 56.5 (#24), GPT-6 Sol: 74.0 (#9)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| ARC-AGI-2 | 61.3% | 89.6% |
| NYT Connections (extended) | 91.3% | 90.1% |
| ARC-AGI-1 | 90.5% | 95.5% |
| CritPt | 18% | 30.9% |
| LMArena Hard Prompts | 1461 | 1418 |
| Mystery Game Puzzles | 43% | 56% |
| DTBench | 93.9% | 97.3% |
| LMCA | 45.5% | 59.1% |
| Epoch Capabilities Index | 155.31 | 162.72 |
| Kagi LLM Benchmark | 53.5% | — |
| Chess Puzzles | 47% | — |
| EBR-Bench | — | 53.3% |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
Math GPT-6 Sol leads
DeepSeek V4 Pro: 64.8 (#30), GPT-6 Sol: 87.2 (#7)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 89.8% |
| FrontierMath Tier 4 | 26.8% | 90% |
| OTIS Mock AIME 2024-2025 | 98.6% | 100% |
| ProofBench | 50% | 83% |
| LMArena Math | 1455 | 1402 |
| MathArena Final-Answer Competitions | 76.6% | — |
Knowledge GPT-6 Sol leads
DeepSeek V4 Pro: 59.5 (#31), GPT-6 Sol: 64.8 (#15)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| GPQA Diamond | 91.7% | 94.3% |
| SimpleQA Verified | 52.9% | 60.7% |
| Vectara Hallucination Rate | 8.6% | 6.5% |
| LMArena Expert | 1464 | 1439 |
Multimodal Not comparable
DeepSeek V4 Pro: —, GPT-6 Sol: 47.6 (#10)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| LMArena Vision | — | 1245 |
| Blueprint-Bench 2 | — | 36.9% |
| Furniture Assembly | — | 58.3% |
Multilingual DeepSeek V4 Pro leads
DeepSeek V4 Pro: 54.4 (#45), GPT-6 Sol: 50.5 (#118)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| LMArena Non-English | 1439 | 1385 |
| LMArena Chinese | 1486 | 1405 |
| LMArena French | 1472 | 1410 |
| LMArena German | 1458 | 1390 |
| LMArena Japanese | 1445 | 1385 |
| LMArena Korean | 1447 | 1341 |
| LMArena Russian | 1453 | 1401 |
| LMArena Spanish | 1458 | 1384 |
Instruction Following DeepSeek V4 Pro leads
DeepSeek V4 Pro: 76.1 (#47), GPT-6 Sol: 74.5 (#94)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| LMArena Instruction Following | 1448 | 1412 |
Long Context DeepSeek V4 Pro leads
DeepSeek V4 Pro: 45.0 (#51), GPT-6 Sol: 43.1 (#108)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| LMArena Longer Query | 1458 | 1411 |
| CL-bench Life | 13.5% | — |
Writing & Preference GPT-6 Sol leads
DeepSeek V4 Pro: 65.5 (#46), GPT-6 Sol: 71.9 (#18)
| Benchmark | DeepSeek V4 Pro | GPT-6 Sol |
|---|---|---|
| LMArena Text | 1451 | 1395 |
| LMArena Creative Writing | 1446 | 1378 |
| EQ-Bench Creative Writing | 1553 | 2125 |
| LMArena Multi-Turn | 1467 | 1412 |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than GPT-6 Sol?
GPT-6 Sol is the stronger model overall, scoring 61.8 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 4.0× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Pro or GPT-6 Sol?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; GPT-6 Sol lists at $2 and $10.
Is DeepSeek V4 Pro or GPT-6 Sol better for coding?
GPT-6 Sol scores higher on coding benchmarks: 60.1 versus 52.4 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Sol does, with 1.05M tokens against 1M.
How many benchmarks do DeepSeek V4 Pro and GPT-6 Sol share?
39 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and GPT-6 Sol has 45.