Model comparison
DeepSeek-V2.5 (Sep 2024) vs GPT-4.1 nano
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 27.9 on the Noometry Index.
Last verified . 15 shared benchmarks.
DeepSeekSummary
- They share 15 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 7 categories and GPT-4.1 nano in 1 category; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek-V2.5 (Sep 2024) leads 25.6 to 8.5.
- The biggest single-benchmark swing is Aider Polyglot: 17.8% for DeepSeek-V2.5 (Sep 2024) and 8.9% for GPT-4.1 nano.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 37.6 | 27.9 |
| Released | 2024-09-06 | 2025-04-14 |
| Weights | Open | Proprietary |
| Context window | — | 1.05M |
| Max output | — | 33K |
| Input $ / M tokens | — | $0.10 |
| Output $ / M tokens | — | $0.40 |
| Results tracked | 22 | 38 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), GPT-4.1 nano: 24.1 (#330)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| Aider Polyglot | 17.8% | 8.9% |
| LMArena Coding | 1309 | 1306 |
| SciCode | — | 25.9% |
| WeirdML | — | 19% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, GPT-4.1 nano: 26.5 (#104)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 33% |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), GPT-4.1 nano: 8.5 (#349)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1286 |
| ARC-AGI-2 | — | 0% |
| Kagi LLM Benchmark | — | 33.3% |
| ARC-AGI-1 | — | 0% |
| CritPt | — | 0% |
| DTBench | — | 52.5% |
| LMCA | — | 5.5% |
| Epoch Capabilities Index | — | 129.62 |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), GPT-4.1 nano: 26.9 (#252)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Math | 1288 | 1274 |
| OTIS Mock AIME 2024-2025 | — | 28.9% |
| Omni-MATH | — | 36.7% |
| MATH Level 5 | — | 70% |
| FrontierMath (Feb 2025 set) | — | 1% |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), GPT-4.1 nano: 21.8 (#273)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Expert | 1266 | 1272 |
| GPQA Diamond | — | 48.9% |
| SimpleQA Verified | — | 6% |
| MMLU-Pro | — | 55% |
| GPQA (HELM) | — | 50.7% |
Multimodal Not comparable
DeepSeek-V2.5 (Sep 2024): —, GPT-4.1 nano: 29.2 (#113)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Vision | — | 1063 |
Multilingual Too close to call
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), GPT-4.1 nano: 41.6 (#205)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Non-English | 1273 | 1260 |
| LMArena Chinese | 1318 | 1270 |
| LMArena German | 1258 | 1288 |
| LMArena Japanese | 1228 | 1198 |
| LMArena Russian | 1289 | 1261 |
| LMArena French | 1289 | — |
| LMArena Korean | 1209 | — |
| LMArena Spanish | 1248 | — |
Instruction Following Too close to call
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), GPT-4.1 nano: 67.8 (#193)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Instruction Following | 1280 | 1267 |
| IFEval | — | 84.3% |
Long Context DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), GPT-4.1 nano: 23.7 (#296)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Longer Query | 1301 | 1283 |
| Fiction.LiveBench | — | 25% |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), GPT-4.1 nano: 40.5 (#243)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | GPT-4.1 nano |
|---|---|---|
| LMArena Text | 1294 | 1285 |
| LMArena Creative Writing | 1285 | 1260 |
| LMArena Multi-Turn | 1297 | 1277 |
| EQ-Bench Creative Writing | — | 946 |
| WildBench | — | 81.2% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than GPT-4.1 nano?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 27.9 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or GPT-4.1 nano better for coding?
DeepSeek-V2.5 (Sep 2024) scores higher on coding benchmarks: 31.7 versus 24.1 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and GPT-4.1 nano share?
15 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and GPT-4.1 nano has 38.