Model comparison
DeepSeek V4 Pro vs GPT-5.6 Terra
GPT-5.6 Terra is the stronger model overall, scoring 59.2 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 4.5× less per token, which makes it the better buy when GPT-5.6 Terra's lead doesn't matter for your workload.
Last verified . 43 shared benchmarks.
DeepSeekSummary
- They share 43 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 1 category and GPT-5.6 Terra in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.6 Terra leads 81.6 to 64.8.
- The biggest single-benchmark swing is FrontierMath Tier 4: 26.8% for DeepSeek V4 Pro and 70.7% for GPT-5.6 Terra.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $2 / $12 for GPT-5.6 Terra.
- GPT-5.6 Terra 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-5.6 Terra | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 54.3 | 59.2 |
| Released | 2026-04-24 | 2026-07-09 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 393K | 128K |
| Input $ / M tokens | $0.66 | $2 |
| Output $ / M tokens | $1.98 | $12 |
| Results tracked | 48 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.6 Terra leads
DeepSeek V4 Pro: 52.4 (#34), GPT-5.6 Terra: 57.7 (#19)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| FrontierCode | 28.6% | 41.3% |
| LMArena WebDev | 1582 | 1522 |
| SciCode | 51% | 55% |
| WeirdML | 66.2% | 78.3% |
| LMArena Coding | 1470 | 1484 |
| ALE-Bench | 1,403 | 1,951 |
| SWE-bench Verified | 77.6% | — |
| DeepSWE | — | 69.6% |
| CursorBench | — | 41.3% |
Agentic & Tool Use GPT-5.6 Terra leads
DeepSeek V4 Pro: 32.8 (#58), GPT-5.6 Terra: 40.1 (#25)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| APEX-Agents | 47.3% | 58.2% |
| Vending-Bench 2 | 3,285 | 7,343 |
| BALROG | — | 53.2% |
| GDP.pdf | — | 24.7% |
Reasoning GPT-5.6 Terra leads
DeepSeek V4 Pro: 56.5 (#24), GPT-5.6 Terra: 60.7 (#21)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| ARC-AGI-2 | 61.3% | 83.9% |
| Kagi LLM Benchmark | 53.5% | 51.3% |
| NYT Connections (extended) | 91.3% | 78.4% |
| ARC-AGI-1 | 90.5% | 96.5% |
| CritPt | 18% | 30% |
| Chess Puzzles | 47% | 54% |
| LMArena Hard Prompts | 1461 | 1468 |
| Mystery Game Puzzles | 43% | 35% |
| DTBench | 93.9% | 93.3% |
| LMCA | 45.5% | 55% |
| Surface Evolver Bench | 40% | 83.8% |
| Epoch Capabilities Index | 155.31 | 159.62 |
| SimpleBench | — | 48.9% |
| ForecastBench | 56.1 | — |
Math GPT-5.6 Terra leads
DeepSeek V4 Pro: 64.8 (#30), GPT-5.6 Terra: 81.6 (#12)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 86% |
| FrontierMath Tier 4 | 26.8% | 70.7% |
| OTIS Mock AIME 2024-2025 | 98.6% | 99.7% |
| ProofBench | 50% | 74% |
| LMArena Math | 1455 | 1466 |
| MathArena Final-Answer Competitions | 76.6% | — |
Knowledge GPT-5.6 Terra leads
DeepSeek V4 Pro: 59.5 (#31), GPT-5.6 Terra: 61.2 (#30)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| GPQA Diamond | 91.7% | 93.3% |
| SimpleQA Verified | 52.9% | 43.2% |
| LMArena Expert | 1464 | 1492 |
| Vectara Hallucination Rate | 8.6% | — |
Multimodal Not comparable
DeepSeek V4 Pro: —, GPT-5.6 Terra: 47.3 (#11)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| LMArena Vision | — | 1271 |
| Blueprint-Bench 2 | — | 30.8% |
| Furniture Assembly | — | 54.2% |
| LMArena Document | — | 1472 |
Multilingual Too close to call
DeepSeek V4 Pro: 54.4 (#45), GPT-5.6 Terra: 54.4 (#44)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| LMArena Non-English | 1439 | 1439 |
| LMArena Chinese | 1486 | 1513 |
| LMArena French | 1472 | 1471 |
| LMArena German | 1458 | 1460 |
| LMArena Japanese | 1445 | 1457 |
| LMArena Korean | 1447 | 1425 |
| LMArena Russian | 1453 | 1450 |
| LMArena Spanish | 1458 | 1448 |
Instruction Following Too close to call
DeepSeek V4 Pro: 76.1 (#47), GPT-5.6 Terra: 76.4 (#40)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| LMArena Instruction Following | 1448 | 1454 |
Long Context Too close to call
DeepSeek V4 Pro: 45.0 (#51), GPT-5.6 Terra: 44.4 (#68)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| LMArena Longer Query | 1458 | 1451 |
| CL-bench Life | 13.5% | — |
Writing & Preference GPT-5.6 Terra leads
DeepSeek V4 Pro: 65.5 (#46), GPT-5.6 Terra: 70.2 (#23)
| Benchmark | DeepSeek V4 Pro | GPT-5.6 Terra |
|---|---|---|
| LMArena Text | 1451 | 1447 |
| LMArena Creative Writing | 1446 | 1410 |
| EQ-Bench Creative Writing | 1553 | 1855 |
| EQ-Bench 4 | 1166 | 1234 |
| LMArena Multi-Turn | 1467 | 1449 |
Frequently asked questions
Is DeepSeek V4 Pro better than GPT-5.6 Terra?
GPT-5.6 Terra is the stronger model overall, scoring 59.2 to 54.3 on the Noometry Index. DeepSeek V4 Pro costs 4.5× less per token, which makes it the better buy when GPT-5.6 Terra's lead doesn't matter for your workload.
Which is cheaper, DeepSeek V4 Pro or GPT-5.6 Terra?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; GPT-5.6 Terra lists at $2 and $12.
Is DeepSeek V4 Pro or GPT-5.6 Terra better for coding?
GPT-5.6 Terra scores higher on coding benchmarks: 57.7 versus 52.4 in the Noometry coding category.
Which has the bigger context window?
GPT-5.6 Terra does, with 1.05M tokens against 1M.
How many benchmarks do DeepSeek V4 Pro and GPT-5.6 Terra share?
43 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and GPT-5.6 Terra has 52.