Model comparison
DeepSeek-V3.1-Terminus vs GPT-5
GPT-5 is the stronger model overall, scoring 50.9 to 43.1 on the Noometry Index. DeepSeek-V3.1-Terminus costs 7.6× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 16 shared benchmarks.
DeepSeekSummary
- They share 16 benchmarks with published results for both. DeepSeek-V3.1-Terminus scores higher in 2 categories and GPT-5 in 5 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 43.4.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 57.4% for DeepSeek-V3.1-Terminus and 72.7% for GPT-5.
- DeepSeek-V3.1-Terminus is cheaper at $0.27 / $1 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 164K.
- DeepSeek-V3.1-Terminus has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.1-Terminus | GPT-5 | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 43.1 | 50.9 |
| Released | 2025-09-22 | 2025-08-07 |
| Weights | Open | Proprietary |
| Context window | 164K | 400K |
| Max output | 147K | 128K |
| Input $ / M tokens | $0.27 | $1.25 |
| Output $ / M tokens | $1 | $10 |
| Results tracked | 16 | 69 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
DeepSeek-V3.1-Terminus: 42.0 (#113), GPT-5: 50.3 (#47)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| SciCode | 40.6% | 42.9% |
| LMArena Coding | 1426 | 1436 |
| ALE-Bench | 745.17 | 1,162 |
| SWE-bench Verified | — | 73.6% |
| SWE-bench Verified (bash only) | — | 65% |
| Aider Polyglot | — | 88% |
| LMArena WebDev | — | 1418 |
| GSO | — | 6.9% |
| WeirdML | — | 60.7% |
| AlgoTune | — | 1.67 |
Agentic & Tool Use Not comparable
DeepSeek-V3.1-Terminus: —, GPT-5: 33.1 (#56)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| Terminal-Bench | — | 49.6% |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| DeepResearch Bench | — | 49.6% |
| BALROG | — | 32.8% |
| LMArena Search | — | 1133 |
| METR Time Horizons | — | 69.6% |
Reasoning GPT-5 leads
DeepSeek-V3.1-Terminus: 26.4 (#133), GPT-5: 38.3 (#64)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| Kagi LLM Benchmark | 57.4% | 72.7% |
| CritPt | 1.7% | 12.6% |
| LMArena Hard Prompts | 1426 | 1416 |
| DTBench | 81.3% | 90.7% |
| LMCA | 28.6% | 40% |
| ARC-AGI-2 | — | 9.9% |
| SimpleBench | — | 56.7% |
| ARC-AGI-1 | — | 65.7% |
| Chess Puzzles | — | 37% |
| EnigmaEval | — | 10.5% |
| EBR-Bench | — | 12.7% |
| Mystery Game Puzzles | — | 23% |
| Epoch Capabilities Index | — | 150 |
| ForecastBench | — | 61.4 |
Math GPT-5 leads
DeepSeek-V3.1-Terminus: 38.5 (#137), GPT-5: 55.0 (#44)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| LMArena Math | 1402 | 1407 |
| FrontierMath (Tiers 1-3) | — | 55.4% |
| FrontierMath Tier 4 | — | 22% |
| OTIS Mock AIME 2024-2025 | — | 91.4% |
| ProofBench | — | 18% |
| Omni-MATH | — | 64.7% |
| MATH Level 5 | — | 98.1% |
| FrontierMath (Feb 2025 set) | — | 32.4% |
| FrontierMath Tier 4 (v1) | — | 12.5% |
Knowledge Not comparable
DeepSeek-V3.1-Terminus: —, GPT-5: 56.6 (#43)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| GPQA Diamond | — | 86.2% |
| Humanity's Last Exam | — | 25.3% |
| SimpleQA Verified | — | 50.1% |
| MMLU-Pro | — | 86.3% |
| Confabulations | — | 10.3% |
| Vectara Hallucination Rate | — | 14.7% |
| GPQA (HELM) | — | 79.2% |
| LMArena Expert | — | 1419 |
Multimodal Not comparable
DeepSeek-V3.1-Terminus: —, GPT-5: 46.8 (#13)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| LMArena Vision | — | 1232 |
| GeoBench | — | 81% |
| VPCT | — | 66% |
Multilingual Too close to call
DeepSeek-V3.1-Terminus: 52.1 (#92), GPT-5: 51.4 (#110)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| LMArena Non-English | 1407 | 1397 |
| LMArena Russian | 1436 | 1406 |
| LMArena Chinese | — | 1422 |
| LMArena French | — | 1410 |
| LMArena German | — | 1416 |
| LMArena Japanese | — | 1409 |
| LMArena Korean | — | 1360 |
| LMArena Spanish | — | 1399 |
Instruction Following Too close to call
DeepSeek-V3.1-Terminus: 74.0 (#106), GPT-5: 73.8 (#113)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| LMArena Instruction Following | 1404 | 1388 |
| IFEval | — | 87.5% |
Long Context GPT-5 leads
DeepSeek-V3.1-Terminus: 43.4 (#97), GPT-5: 69.5 (#2)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| LMArena Longer Query | 1421 | 1399 |
| Fiction.LiveBench | — | 97.2% |
Writing & Preference GPT-5 leads
DeepSeek-V3.1-Terminus: 61.0 (#92), GPT-5: 63.4 (#65)
| Benchmark | DeepSeek-V3.1-Terminus | GPT-5 |
|---|---|---|
| LMArena Text | 1419 | 1406 |
| LMArena Creative Writing | 1403 | 1365 |
| LMArena Multi-Turn | 1411 | 1426 |
| Short-Story Creative Writing | — | 86% |
| EQ-Bench Creative Writing | — | 1627 |
| WildBench | — | 85.7% |
Frequently asked questions
Is DeepSeek-V3.1-Terminus better than GPT-5?
GPT-5 is the stronger model overall, scoring 50.9 to 43.1 on the Noometry Index. DeepSeek-V3.1-Terminus costs 7.6× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.1-Terminus or GPT-5?
DeepSeek-V3.1-Terminus is cheaper. It lists at $0.27 per million input tokens and $1 per million output tokens; GPT-5 lists at $1.25 and $10.
Is DeepSeek-V3.1-Terminus or GPT-5 better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 42.0 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 164K.
How many benchmarks do DeepSeek-V3.1-Terminus and GPT-5 share?
16 benchmarks have published results for both models. DeepSeek-V3.1-Terminus has 16 scored results on Noometry and GPT-5 has 69.