Model comparison
DeepSeek-V2.5 (Sep 2024) vs GPT-6 Sol
GPT-6 Sol is the stronger model overall, scoring 61.8 to 37.6 on the Noometry Index.
Last verified . 17 shared benchmarks.
DeepSeekSummary
- They share 17 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 0 categories and GPT-6 Sol in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Sol leads 87.2 to 35.9.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 37.6 | 61.8 |
| Released | 2024-09-06 | 2026-09-22 |
| Weights | Open | Proprietary |
| Context window | — | 1.05M |
| Max output | — | 128K |
| Input $ / M tokens | — | $2 |
| Output $ / M tokens | — | $10 |
| Results tracked | 22 | 45 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), GPT-6 Sol: 60.1 (#11)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Coding | 1309 | 1447 |
| DeepSWE | — | 68.8% |
| FrontierCode | — | 49.3% |
| Aider Polyglot | 17.8% | — |
| LMArena WebDev | — | 1688 |
| SciCode | — | 57.6% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 2,462 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, GPT-6 Sol: 37.2 (#36)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| APEX-Agents | — | 54.3% |
| GDP.pdf | — | 26.4% |
| Vending-Bench 2 | — | 14,428 |
Reasoning GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), GPT-6 Sol: 74.0 (#9)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1418 |
| ARC-AGI-2 | — | 89.6% |
| NYT Connections (extended) | — | 90.1% |
| ARC-AGI-1 | — | 95.5% |
| CritPt | — | 30.9% |
| EBR-Bench | — | 53.3% |
| Mystery Game Puzzles | — | 56% |
| DTBench | — | 97.3% |
| LMCA | — | 59.1% |
| Epoch Capabilities Index | — | 162.72 |
Math GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), GPT-6 Sol: 87.2 (#7)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Math | 1288 | 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-V2.5 (Sep 2024): 34.8 (#193), GPT-6 Sol: 64.8 (#15)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Expert | 1266 | 1439 |
| GPQA Diamond | — | 94.3% |
| SimpleQA Verified | — | 60.7% |
| Vectara Hallucination Rate | — | 6.5% |
Multimodal Not comparable
DeepSeek-V2.5 (Sep 2024): —, GPT-6 Sol: 47.6 (#10)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Vision | — | 1245 |
| Blueprint-Bench 2 | — | 36.9% |
| Furniture Assembly | — | 58.3% |
Multilingual GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), GPT-6 Sol: 50.5 (#118)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Non-English | 1273 | 1385 |
| LMArena Chinese | 1318 | 1405 |
| LMArena French | 1289 | 1410 |
| LMArena German | 1258 | 1390 |
| LMArena Japanese | 1228 | 1385 |
| LMArena Korean | 1209 | 1341 |
| LMArena Russian | 1289 | 1401 |
| LMArena Spanish | 1248 | 1384 |
Instruction Following GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), GPT-6 Sol: 74.5 (#94)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Instruction Following | 1280 | 1412 |
Long Context GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), GPT-6 Sol: 43.1 (#108)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Longer Query | 1301 | 1411 |
Writing & Preference GPT-6 Sol leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), GPT-6 Sol: 71.9 (#18)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-6 Sol |
|---|---|---|
| LMArena Text | 1294 | 1395 |
| LMArena Creative Writing | 1285 | 1378 |
| LMArena Multi-Turn | 1297 | 1412 |
| EQ-Bench Creative Writing | — | 2125 |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than GPT-6 Sol?
GPT-6 Sol is the stronger model overall, scoring 61.8 to 37.6 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or GPT-6 Sol better for coding?
GPT-6 Sol scores higher on coding benchmarks: 60.1 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and GPT-6 Sol share?
17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and GPT-6 Sol has 45.