Model comparison
DeepSeek-R1 vs GPT-5 Nano
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 33.5 on the Noometry Index. GPT-5 Nano costs 6.6× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.
Last verified . 34 shared benchmarks.
DeepSeekSummary
- They share 34 benchmarks with published results for both. DeepSeek-R1 scores higher in 8 categories and GPT-5 Nano in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-R1 leads 61.4 to 39.1.
- The biggest single-benchmark swing is Fiction.LiveBench: 75% for DeepSeek-R1 and 44.4% for GPT-5 Nano.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $0.50 / $2.15 for DeepSeek-R1.
- GPT-5 Nano accepts more context: 400K tokens versus 164K.
Side by side
| DeepSeek-R1 | GPT-5 Nano | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 42.3 | 33.5 |
| Released | 2025-01-20 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 164K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $0.50 | $0.05 |
| Output $ / M tokens | $2.15 | $0.40 |
| Results tracked | 52 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 46.3 (#68), GPT-5 Nano: 33.6 (#254)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| WeirdML | 41.6% | 38.1% |
| LMArena Coding | 1427 | 1351 |
| ALE-Bench | 804.12 | 718.67 |
| SWE-bench Verified (bash only) | — | 34.8% |
| Aider Polyglot | 71.4% | — |
| SciCode | 35.7% | — |
| LiveBench Coding | 66.7% | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use DeepSeek-R1 leads
DeepSeek-R1: 30.7 (#75), GPT-5 Nano: 25.8 (#106)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| Terminal-Bench | — | 21.8% |
| Berkeley Function Calling Leaderboard | — | 51.5% |
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning DeepSeek-R1 leads
DeepSeek-R1: 18.6 (#278), GPT-5 Nano: 16.3 (#306)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| ARC-AGI-2 | 1.3% | 2.6% |
| Kagi LLM Benchmark | 69.4% | 62.2% |
| ARC-AGI-1 | 21.2% | 20.7% |
| LMArena Hard Prompts | 1416 | 1328 |
| Epoch Capabilities Index | 141.29 | 139.38 |
| ForecastBench | 60 | 59.1 |
| SimpleBench | 40.8% | — |
| CritPt | 1.1% | — |
| Chess Puzzles | — | 27% |
| LiveBench Reasoning | 83.2% | — |
| Mystery Game Puzzles | — | 9% |
| DTBench | — | 62.7% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 7.9% |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 43.8 (#79), GPT-5 Nano: 29.4 (#241)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 81.1% |
| Omni-MATH | 42.4% | 54.6% |
| LMArena Math | 1400 | 1317 |
| MATH Level 5 | 96.6% | 95.2% |
| FrontierMath (Tiers 1-3) | — | 20% |
| FrontierMath Tier 4 | — | 2.4% |
| ProofBench | — | 12% |
| LiveBench Math | 80.7% | — |
| FrontierMath (Feb 2025 set) | — | 8.3% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.5 (#87), GPT-5 Nano: 35.9 (#178)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| GPQA Diamond | 76.3% | 69.4% |
| MMLU-Pro | 79.3% | 77.8% |
| Vectara Hallucination Rate | 11.3% | 10.5% |
| GPQA (HELM) | 66.6% | 67.9% |
| LMArena Expert | 1394 | 1321 |
| SimpleQA Verified | — | 11.7% |
| Confabulations | 12.7% | — |
Multimodal Not comparable
DeepSeek-R1: —, GPT-5 Nano: 31.3 (#108)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| LMArena Vision | — | 1159 |
| VPCT | — | 37.2% |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.4 (#85), GPT-5 Nano: 45.3 (#172)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| LMArena Non-English | 1412 | 1313 |
| LMArena Chinese | 1442 | 1356 |
| LMArena German | 1404 | 1327 |
| LMArena Japanese | 1391 | 1226 |
| LMArena Korean | 1360 | 1269 |
| LMArena Russian | 1423 | 1296 |
| LMArena Spanish | 1411 | 1360 |
| LMArena French | 1417 | — |
Instruction Following GPT-5 Nano leads
DeepSeek-R1: 72.0 (#143), GPT-5 Nano: 75.0 (#79)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| IFEval | 78.4% | 93.2% |
| LMArena Instruction Following | 1382 | 1306 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), GPT-5 Nano: 31.3 (#281)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| Fiction.LiveBench | 75% | 44.4% |
| LMArena Longer Query | 1391 | 1312 |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.4 (#88), GPT-5 Nano: 39.1 (#249)
| Benchmark | DeepSeek-R1 | GPT-5 Nano |
|---|---|---|
| LMArena Text | 1428 | 1320 |
| LMArena Creative Writing | 1405 | 1249 |
| EQ-Bench Creative Writing | 1500 | 705 |
| WildBench | 82.8% | 80.6% |
| LMArena Multi-Turn | 1405 | 1311 |
| Short-Story Creative Writing | 83% | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than GPT-5 Nano?
DeepSeek-R1 is the stronger model overall, scoring 42.3 to 33.5 on the Noometry Index. GPT-5 Nano costs 6.6× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-R1 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-R1 lists at $0.50 and $2.15.
Is DeepSeek-R1 or GPT-5 Nano better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 46.3 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-R1 and GPT-5 Nano share?
34 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GPT-5 Nano has 49.