Model comparison
DeepSeek V4 Pro vs GPT-5.5
GPT-5.5 is the stronger model overall, scoring 63.4 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 11× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.
Last verified . 48 shared benchmarks.
DeepSeekSummary
- They share 48 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 0 categories and GPT-5.5 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where GPT-5.5 leads 50.7 to 32.8.
- The biggest single-benchmark swing is Surface Evolver Bench: 40% for DeepSeek V4 Pro and 88.1% for GPT-5.5.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $5 / $30 for GPT-5.5.
- GPT-5.5 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-5.5 | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 54.3 | 63.4 |
| Released | 2026-04-24 | 2026-04-23 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 393K | 128K |
| Input $ / M tokens | $0.66 | $5 |
| Output $ / M tokens | $1.98 | $30 |
| Results tracked | 48 | 71 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.5 leads
DeepSeek V4 Pro: 52.4 (#34), GPT-5.5: 58.2 (#17)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| SWE-bench Verified | 77.6% | 80.6% |
| FrontierCode | 28.6% | 43% |
| LMArena WebDev | 1582 | 1513 |
| SciCode | 51% | 56.1% |
| WeirdML | 66.2% | 84.9% |
| LMArena Coding | 1470 | 1494 |
| ALE-Bench | 1,403 | 1,943 |
| DeepSWE | — | 67% |
| GSO | — | 40.2% |
| MirrorCode | — | 10% |
Agentic & Tool Use GPT-5.5 leads
DeepSeek V4 Pro: 32.8 (#58), GPT-5.5: 50.7 (#6)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| APEX-Agents | 47.3% | 55.1% |
| Vending-Bench 2 | 3,285 | 7,524 |
| Terminal-Bench | — | 84.7% |
| OSWorld 2.0 | — | 13% |
| Remote Labor Index | — | 6.3% |
| τ²-bench Banking | — | 44.6% |
| DeepResearch Bench | — | 54% |
| PostTrainBench | — | 27.2% |
| ExploitBench | — | 47.4% |
| GBAEval | — | 53.2% |
| GDP.pdf | — | 26% |
| LMArena Search | — | 1242 |
Reasoning GPT-5.5 leads
DeepSeek V4 Pro: 56.5 (#24), GPT-5.5: 72.8 (#11)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| ARC-AGI-2 | 61.3% | 85% |
| Kagi LLM Benchmark | 53.5% | 88.8% |
| NYT Connections (extended) | 91.3% | 96.2% |
| ARC-AGI-1 | 90.5% | 95% |
| CritPt | 18% | 27.1% |
| Chess Puzzles | 47% | 54% |
| LMArena Hard Prompts | 1461 | 1489 |
| Mystery Game Puzzles | 43% | 56% |
| DTBench | 93.9% | 96% |
| LMCA | 45.5% | 54.3% |
| Surface Evolver Bench | 40% | 88.1% |
| Epoch Capabilities Index | 155.31 | 159.1 |
| ForecastBench | 56.1 | 60.6 |
| SimpleBench | — | 69% |
| EBR-Bench | — | 34.3% |
| Bench to the Future 3 | — | 0.14 |
Math GPT-5.5 leads
DeepSeek V4 Pro: 64.8 (#30), GPT-5.5: 81.7 (#11)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 85.3% |
| FrontierMath Tier 4 | 26.8% | 72.5% |
| MathArena Final-Answer Competitions | 76.6% | 94.3% |
| OTIS Mock AIME 2024-2025 | 98.6% | 100% |
| ProofBench | 50% | 50% |
| LMArena Math | 1455 | 1486 |
| FrontierMath (Feb 2025 set) | — | 51.7% |
| FrontierMath Erdős | — | 0% |
| FrontierMath Tier 4 (v1) | — | 35.4% |
Knowledge GPT-5.5 leads
DeepSeek V4 Pro: 59.5 (#31), GPT-5.5: 64.4 (#17)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| GPQA Diamond | 91.7% | 94% |
| SimpleQA Verified | 52.9% | 63% |
| Vectara Hallucination Rate | 8.6% | 9.3% |
| LMArena Expert | 1464 | 1508 |
Multimodal Not comparable
DeepSeek V4 Pro: —, GPT-5.5: 46.9 (#12)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| LMArena Vision | — | 1297 |
| Blueprint-Bench 2 | — | 36.2% |
| Furniture Assembly | — | 44.2% |
| LMArena Document | — | 1486 |
Multilingual GPT-5.5 leads
DeepSeek V4 Pro: 54.4 (#45), GPT-5.5: 56.4 (#20)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| LMArena Non-English | 1439 | 1467 |
| LMArena Chinese | 1486 | 1533 |
| LMArena French | 1472 | 1486 |
| LMArena German | 1458 | 1480 |
| LMArena Japanese | 1445 | 1498 |
| LMArena Korean | 1447 | 1460 |
| LMArena Russian | 1453 | 1473 |
| LMArena Spanish | 1458 | 1468 |
Instruction Following GPT-5.5 leads
DeepSeek V4 Pro: 76.1 (#47), GPT-5.5: 77.5 (#18)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| LMArena Instruction Following | 1448 | 1479 |
Long Context GPT-5.5 leads
DeepSeek V4 Pro: 45.0 (#51), GPT-5.5: 48.3 (#12)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| CL-bench Life | 13.5% | 22.2% |
| LMArena Longer Query | 1458 | 1484 |
Writing & Preference GPT-5.5 leads
DeepSeek V4 Pro: 65.5 (#46), GPT-5.5: 72.7 (#13)
| Benchmark | DeepSeek V4 Pro | GPT-5.5 |
|---|---|---|
| LMArena Text | 1451 | 1472 |
| LMArena Creative Writing | 1446 | 1455 |
| EQ-Bench Creative Writing | 1553 | 1844 |
| EQ-Bench 4 | 1166 | 1315 |
| LMArena Multi-Turn | 1467 | 1476 |
Frequently asked questions
Is DeepSeek V4 Pro better than GPT-5.5?
GPT-5.5 is the stronger model overall, scoring 63.4 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 11× less per token, which makes it the better buy when GPT-5.5's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Pro or GPT-5.5?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; GPT-5.5 lists at $5 and $30.
Is DeepSeek V4 Pro or GPT-5.5 better for coding?
GPT-5.5 scores higher on coding benchmarks: 58.2 versus 52.4 in the Noometry coding category.
Which has the bigger context window?
GPT-5.5 does, with 1.05M tokens against 1M.
How many benchmarks do DeepSeek V4 Pro and GPT-5.5 share?
48 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and GPT-5.5 has 71.