Model comparison
DeepSeek-V3.1 vs GPT-5 Nano
DeepSeek-V3.1 is the stronger model overall, scoring 42.2 to 33.4 on the Noometry Index. GPT-5 Nano costs 3.1× less per token, which makes it the better buy when DeepSeek-V3.1's lead doesn't matter for your workload.
Last verified . 24 shared benchmarks.
DeepSeekSummary
- They share 24 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 7 categories and GPT-5 Nano in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.1 leads 60.5 to 39.4.
- The biggest single-benchmark swing is DTBench: 82.7% for DeepSeek-V3.1 and 62.7% for GPT-5 Nano.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $0.25 / $0.95 for DeepSeek-V3.1.
- GPT-5 Nano accepts more context: 400K tokens versus 164K.
- DeepSeek-V3.1 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.1 | GPT-5 Nano | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.2 | 33.4 |
| Released | 2025-08-21 | 2025-08-07 |
| Weights | Open | Proprietary |
| Context window | 164K | 400K |
| Max output | 8K | 128K |
| Input $ / M tokens | $0.25 | $0.05 |
| Output $ / M tokens | $0.95 | $0.40 |
| Results tracked | 27 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1 leads
DeepSeek-V3.1: 39.8 (#140), GPT-5 Nano: 33.3 (#248)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| WeirdML | 38.4% | 38.1% |
| LMArena Coding | 1417 | 1351 |
| SWE-bench Verified (bash only) | — | 34.8% |
| ALE-Bench | — | 718.67 |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, GPT-5 Nano: 25.6 (#99)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| Terminal-Bench | — | 21.8% |
| Berkeley Function Calling Leaderboard | — | 51.5% |
Reasoning DeepSeek-V3.1 leads
DeepSeek-V3.1: 27.3 (#109), GPT-5 Nano: 16.4 (#291)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| Kagi LLM Benchmark | 53.2% | 62.2% |
| LMArena Hard Prompts | 1417 | 1329 |
| DTBench | 82.7% | 62.7% |
| LMCA | 24.3% | 7.9% |
| Epoch Capabilities Index | 139.92 | 139.39 |
| ForecastBench | 58 | 59.1 |
| ARC-AGI-2 | — | 2.6% |
| SimpleBench | 40% | — |
| ARC-AGI-1 | — | 20.7% |
| Chess Puzzles | — | 27% |
| Mystery Game Puzzles | — | 9% |
Math DeepSeek-V3.1 leads
DeepSeek-V3.1: 36.1 (#125), GPT-5 Nano: 29.1 (#221)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| LMArena Math | 1420 | 1318 |
| FrontierMath (Tiers 1-3) | — | 20% |
| FrontierMath Tier 4 | — | 2.4% |
| OTIS Mock AIME 2024-2025 | — | 81.1% |
| ProofBench | — | 12% |
| Omni-MATH | — | 54.6% |
| MATH Level 5 | — | 95.2% |
| FrontierMath (Feb 2025 set) | — | 8.3% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge DeepSeek-V3.1 leads
DeepSeek-V3.1: 43.0 (#86), GPT-5 Nano: 35.8 (#166)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| Vectara Hallucination Rate | 5.5% | 10.5% |
| LMArena Expert | 1403 | 1321 |
| GPQA Diamond | — | 69.4% |
| SimpleQA Verified | — | 11.7% |
| MMLU-Pro | — | 77.8% |
| GPQA (HELM) | — | 67.9% |
Multimodal Not comparable
DeepSeek-V3.1: —, GPT-5 Nano: 30.7 (#102)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| LMArena Vision | — | 1159 |
| VPCT | — | 37.2% |
Multilingual DeepSeek-V3.1 leads
DeepSeek-V3.1: 51.8 (#100), GPT-5 Nano: 45.5 (#167)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| LMArena Non-English | 1400 | 1313 |
| LMArena Chinese | 1470 | 1355 |
| LMArena German | 1412 | 1329 |
| LMArena Korean | 1337 | 1269 |
| LMArena Russian | 1405 | 1296 |
| LMArena Spanish | 1429 | 1360 |
| LMArena French | 1447 | — |
| LMArena Japanese | 1374 | — |
Instruction Following GPT-5 Nano leads
DeepSeek-V3.1: 74.1 (#105), GPT-5 Nano: 75.2 (#72)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| LMArena Instruction Following | 1400 | 1306 |
| IFEval | — | 93.2% |
Long Context DeepSeek-V3.1 leads
DeepSeek-V3.1: 36.3 (#227), GPT-5 Nano: 31.3 (#275)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| Fiction.LiveBench | 52.8% | 44.4% |
| LMArena Longer Query | 1422 | 1313 |
Writing & Preference DeepSeek-V3.1 leads
DeepSeek-V3.1: 60.5 (#93), GPT-5 Nano: 39.4 (#244)
| Benchmark | DeepSeek-V3.1 | GPT-5 Nano |
|---|---|---|
| LMArena Text | 1420 | 1320 |
| LMArena Creative Writing | 1401 | 1250 |
| EQ-Bench Creative Writing | 1436 | 705 |
| LMArena Multi-Turn | 1408 | 1311 |
| WildBench | — | 80.6% |
Frequently asked questions
Is DeepSeek-V3.1 better than GPT-5 Nano?
DeepSeek-V3.1 is the stronger model overall, scoring 42.2 to 33.4 on the Noometry Index. GPT-5 Nano costs 3.1× less per token, which makes it the better buy when DeepSeek-V3.1's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.1 or GPT-5 Nano?
GPT-5 Nano is cheaper. It lists at $0.05 per million input tokens and $0.40 per million output tokens; DeepSeek-V3.1 lists at $0.25 and $0.95.
Is DeepSeek-V3.1 or GPT-5 Nano better for coding?
DeepSeek-V3.1 scores higher on coding benchmarks: 39.8 versus 33.3 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Nano does, with 400K tokens against 164K.
How many benchmarks do DeepSeek-V3.1 and GPT-5 Nano share?
24 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and GPT-5 Nano has 48.