Model comparison
DeepSeek-V3.1 vs GPT-6 Sol
GPT-6 Sol is the stronger model overall, scoring 61.8 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 9.4× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
DeepSeekSummary
- They share 22 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 1 category and GPT-6 Sol in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Sol leads 87.2 to 38.9.
- The biggest single-benchmark swing is LMCA: 24.3% for DeepSeek-V3.1 and 59.1% for GPT-6 Sol.
- DeepSeek-V3.1 is cheaper at $0.25 / $0.95 per million input/output tokens, against $2 / $10 for GPT-6 Sol.
- GPT-6 Sol accepts more context: 1.05M tokens versus 164K.
- DeepSeek-V3.1 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.1 | GPT-6 Sol | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.8 | 61.8 |
| Released | 2025-08-21 | 2026-09-22 |
| Weights | Open | Proprietary |
| Context window | 164K | 1.05M |
| Max output | 8K | 128K |
| Input $ / M tokens | $0.25 | $2 |
| Output $ / M tokens | $0.95 | $10 |
| Results tracked | 27 | 45 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Sol leads
DeepSeek-V3.1: 40.3 (#144), GPT-6 Sol: 60.1 (#11)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Coding | 1417 | 1447 |
| DeepSWE | — | 68.8% |
| FrontierCode | — | 49.3% |
| LMArena WebDev | — | 1688 |
| SciCode | — | 57.6% |
| WeirdML | 38.4% | — |
| ALE-Bench | — | 2,462 |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, GPT-6 Sol: 37.2 (#36)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| APEX-Agents | — | 54.3% |
| GDP.pdf | — | 26.4% |
| Vending-Bench 2 | — | 14,428 |
Reasoning GPT-6 Sol leads
DeepSeek-V3.1: 27.9 (#110), GPT-6 Sol: 74.0 (#9)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Hard Prompts | 1417 | 1418 |
| DTBench | 82.7% | 97.3% |
| LMCA | 24.3% | 59.1% |
| Epoch Capabilities Index | 139.92 | 162.72 |
| ARC-AGI-2 | — | 89.6% |
| SimpleBench | 40% | — |
| Kagi LLM Benchmark | 53.2% | — |
| NYT Connections (extended) | — | 90.1% |
| ARC-AGI-1 | — | 95.5% |
| CritPt | — | 30.9% |
| EBR-Bench | — | 53.3% |
| Mystery Game Puzzles | — | 56% |
| ForecastBench | 58 | — |
Math GPT-6 Sol leads
DeepSeek-V3.1: 38.9 (#122), GPT-6 Sol: 87.2 (#7)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Math | 1420 | 1402 |
| FrontierMath (Tiers 1-3) | — | 89.8% |
| FrontierMath Tier 4 | — | 90% |
| OTIS Mock AIME 2024-2025 | — | 100% |
| ProofBench | — | 83% |
Knowledge GPT-6 Sol leads
DeepSeek-V3.1: 43.7 (#90), GPT-6 Sol: 64.8 (#15)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| Vectara Hallucination Rate | 5.5% | 6.5% |
| LMArena Expert | 1405 | 1439 |
| GPQA Diamond | — | 94.3% |
| SimpleQA Verified | — | 60.7% |
Multimodal Not comparable
DeepSeek-V3.1: —, GPT-6 Sol: 47.6 (#10)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Vision | — | 1245 |
| Blueprint-Bench 2 | — | 36.9% |
| Furniture Assembly | — | 58.3% |
Multilingual DeepSeek-V3.1 leads
DeepSeek-V3.1: 51.6 (#106), GPT-6 Sol: 50.5 (#118)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Non-English | 1400 | 1385 |
| LMArena Chinese | 1469 | 1405 |
| LMArena French | 1447 | 1410 |
| LMArena German | 1411 | 1390 |
| LMArena Japanese | 1378 | 1385 |
| LMArena Korean | 1337 | 1341 |
| LMArena Russian | 1405 | 1401 |
| LMArena Spanish | 1431 | 1384 |
Instruction Following Too close to call
DeepSeek-V3.1: 73.9 (#110), GPT-6 Sol: 74.5 (#94)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Instruction Following | 1400 | 1412 |
Long Context GPT-6 Sol leads
DeepSeek-V3.1: 36.3 (#232), GPT-6 Sol: 43.1 (#108)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Longer Query | 1422 | 1411 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference GPT-6 Sol leads
DeepSeek-V3.1: 60.3 (#98), GPT-6 Sol: 71.9 (#18)
| Benchmark | DeepSeek-V3.1 | GPT-6 Sol |
|---|---|---|
| LMArena Text | 1420 | 1395 |
| LMArena Creative Writing | 1401 | 1378 |
| EQ-Bench Creative Writing | 1436 | 2125 |
| LMArena Multi-Turn | 1408 | 1412 |
Frequently asked questions
Is DeepSeek-V3.1 better than GPT-6 Sol?
GPT-6 Sol is the stronger model overall, scoring 61.8 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 9.4× 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-V3.1 or GPT-6 Sol?
DeepSeek-V3.1 is cheaper. It lists at $0.25 per million input tokens and $0.95 per million output tokens; GPT-6 Sol lists at $2 and $10.
Is DeepSeek-V3.1 or GPT-6 Sol better for coding?
GPT-6 Sol scores higher on coding benchmarks: 60.1 versus 40.3 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Sol does, with 1.05M tokens against 164K.
How many benchmarks do DeepSeek-V3.1 and GPT-6 Sol share?
22 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and GPT-6 Sol has 45.