Model comparison
DeepSeek-V3.2-Exp vs GPT-5 Nano
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 33.5 on the Noometry Index. GPT-5 Nano costs 2.1× less per token, which makes it the better buy when DeepSeek-V3.2-Exp's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
DeepSeekSummary
- They share 35 benchmarks with published results for both. DeepSeek-V3.2-Exp scores higher in 8 categories and GPT-5 Nano in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.2-Exp leads 62.4 to 39.1.
- The biggest single-benchmark swing is Fiction.LiveBench: 83.3% for DeepSeek-V3.2-Exp and 44.4% for GPT-5 Nano.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $0.26 / $0.38 for DeepSeek-V3.2-Exp.
- GPT-5 Nano accepts more context: 400K tokens versus 164K.
- DeepSeek-V3.2-Exp has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.2-Exp | GPT-5 Nano | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 44.3 | 33.5 |
| Released | 2025-09-29 | 2025-08-07 |
| Weights | Open | Proprietary |
| Context window | 164K | 400K |
| Max output | 66K | 128K |
| Input $ / M tokens | $0.26 | $0.05 |
| Output $ / M tokens | $0.38 | $0.40 |
| Results tracked | 49 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 46.5 (#65), GPT-5 Nano: 33.6 (#254)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| SWE-bench Verified (bash only) | 70% | 34.8% |
| WeirdML | 39.5% | 38.1% |
| LMArena Coding | 1454 | 1351 |
| Aider Polyglot | 74.2% | — |
| LMArena WebDev | 1362 | — |
| SWE-bench Multilingual | 59% | — |
| SciCode | 38.9% | — |
| ALE-Bench | — | 718.67 |
Agentic & Tool Use DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 32.7 (#59), GPT-5 Nano: 25.8 (#106)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| Terminal-Bench | 39.6% | 21.8% |
| Berkeley Function Calling Leaderboard | 56.7% | 51.5% |
| APEX-Agents | 21.3% | — |
| TheAgentCompany | 42.9% | — |
| Vending-Bench 2 | 1,034 | — |
Reasoning DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 22.1 (#208), GPT-5 Nano: 16.3 (#306)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| ARC-AGI-2 | 4% | 2.6% |
| Kagi LLM Benchmark | 52.2% | 62.2% |
| ARC-AGI-1 | 57% | 20.7% |
| Chess Puzzles | 14% | 27% |
| LMArena Hard Prompts | 1434 | 1328 |
| DTBench | 87.7% | 62.7% |
| LMCA | 29.1% | 7.9% |
| Epoch Capabilities Index | 146.27 | 139.38 |
| NYT Connections (extended) | 36.7% | — |
| CritPt | 2.9% | — |
| Thematic Generalization | 65% | — |
| Mystery Game Puzzles | — | 9% |
| ForecastBench | — | 59.1 |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 41.7 (#87), GPT-5 Nano: 29.4 (#241)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 87.8% | 81.1% |
| ProofBench | 8% | 12% |
| LMArena Math | 1435 | 1317 |
| FrontierMath (Feb 2025 set) | 22.1% | 8.3% |
| FrontierMath Tier 4 (v1) | 2.1% | 2.1% |
| FrontierMath (Tiers 1-3) | — | 20% |
| FrontierMath Tier 4 | — | 2.4% |
| MathArena Final-Answer Competitions | 57.7% | — |
| Omni-MATH | — | 54.6% |
| MATH Level 5 | — | 95.2% |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.7 (#66), GPT-5 Nano: 35.9 (#178)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| GPQA Diamond | 83.4% | 69.4% |
| Vectara Hallucination Rate | 5.3% | 10.5% |
| LMArena Expert | 1436 | 1321 |
| SimpleQA Verified | — | 11.7% |
| MMLU-Pro | — | 77.8% |
| GPQA (HELM) | — | 67.9% |
Multimodal Not comparable
DeepSeek-V3.2-Exp: —, GPT-5 Nano: 31.3 (#108)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| LMArena Vision | — | 1159 |
| VPCT | — | 37.2% |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 52.2 (#90), GPT-5 Nano: 45.3 (#172)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| LMArena Non-English | 1409 | 1313 |
| LMArena Chinese | 1461 | 1356 |
| LMArena German | 1440 | 1327 |
| LMArena Japanese | 1374 | 1226 |
| LMArena Korean | 1371 | 1269 |
| LMArena Russian | 1424 | 1296 |
| LMArena Spanish | 1440 | 1360 |
| LMArena French | 1433 | — |
Instruction Following Too close to call
DeepSeek-V3.2-Exp: 74.5 (#93), GPT-5 Nano: 75.0 (#79)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| LMArena Instruction Following | 1413 | 1306 |
| IFEval | — | 93.2% |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#16), GPT-5 Nano: 31.3 (#281)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| Fiction.LiveBench | 83.3% | 44.4% |
| LMArena Longer Query | 1428 | 1312 |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.4 (#77), GPT-5 Nano: 39.1 (#249)
| Benchmark | DeepSeek-V3.2-Exp | GPT-5 Nano |
|---|---|---|
| LMArena Text | 1425 | 1320 |
| LMArena Creative Writing | 1403 | 1249 |
| EQ-Bench Creative Writing | 1515 | 705 |
| LMArena Multi-Turn | 1427 | 1311 |
| WildBench | — | 80.6% |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than GPT-5 Nano?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 33.5 on the Noometry Index. GPT-5 Nano costs 2.1× less per token, which makes it the better buy when DeepSeek-V3.2-Exp's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.2-Exp 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-V3.2-Exp lists at $0.26 and $0.38.
Is DeepSeek-V3.2-Exp or GPT-5 Nano better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 versus 33.6 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-V3.2-Exp and GPT-5 Nano share?
35 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and GPT-5 Nano has 49.