Model comparison
GPT-4 vs GPT-5 Nano
GPT-5 Nano is the stronger model overall, scoring 33.5 to 29.1 on the Noometry Index.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. GPT-4 scores higher in 2 categories and GPT-5 Nano in 6 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5 Nano leads 29.4 to 10.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.1% for GPT-4 and 81.1% for GPT-5 Nano.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $30 / $60 for GPT-4.
- GPT-5 Nano accepts more context: 400K tokens versus 8K.
Side by side
| GPT-4 | GPT-5 Nano | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 29.1 | 33.5 |
| Released | 2023-03-14 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 8K | 400K |
| Max output | 8K | 128K |
| Input $ / M tokens | $30 | $0.05 |
| Output $ / M tokens | $60 | $0.40 |
| Results tracked | 38 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 Nano leads
GPT-4: 31.6 (#283), GPT-5 Nano: 33.6 (#254)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| WeirdML | 12.4% | 38.1% |
| LMArena Coding | 1254 | 1351 |
| SWE-bench Verified (bash only) | — | 34.8% |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| ALE-Bench | — | 718.67 |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, GPT-5 Nano: 25.8 (#106)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| Terminal-Bench | — | 21.8% |
| Berkeley Function Calling Leaderboard | — | 51.5% |
| METR Time Horizons | 36.1% | — |
Reasoning GPT-4 leads
GPT-4: 17.8 (#289), GPT-5 Nano: 16.3 (#306)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| Chess Puzzles | 4% | 27% |
| LMArena Hard Prompts | 1241 | 1328 |
| Mystery Game Puzzles | 12% | 9% |
| DTBench | 62.7% | 62.7% |
| LMCA | 17.1% | 7.9% |
| Epoch Capabilities Index | 125.89 | 139.38 |
| ForecastBench | 57.8 | 59.1 |
| ARC-AGI-2 | — | 2.6% |
| Kagi LLM Benchmark | — | 62.2% |
| ARC-AGI-1 | — | 20.7% |
| BIG-Bench Hard | 75.1% | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math GPT-5 Nano leads
GPT-4: 10.8 (#309), GPT-5 Nano: 29.4 (#241)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 81.1% |
| LMArena Math | 1269 | 1317 |
| MATH Level 5 | 23% | 95.2% |
| FrontierMath (Tiers 1-3) | — | 20% |
| FrontierMath Tier 4 | — | 2.4% |
| ProofBench | — | 12% |
| Omni-MATH | — | 54.6% |
| FrontierMath (Feb 2025 set) | — | 8.3% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
| GSM8K | 92% | — |
Knowledge GPT-5 Nano leads
GPT-4: 18.4 (#282), GPT-5 Nano: 35.9 (#178)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| GPQA Diamond | 35.7% | 69.4% |
| LMArena Expert | 1211 | 1321 |
| SimpleQA Verified | — | 11.7% |
| MMLU-Pro | — | 77.8% |
| Vectara Hallucination Rate | — | 10.5% |
| GPQA (HELM) | — | 67.9% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, GPT-5 Nano: 31.3 (#108)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| LMArena Vision | — | 1159 |
| VPCT | — | 37.2% |
Multilingual GPT-5 Nano leads
GPT-4: 40.6 (#215), GPT-5 Nano: 45.3 (#172)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| LMArena Non-English | 1246 | 1313 |
| LMArena Chinese | 1242 | 1356 |
| LMArena German | 1251 | 1327 |
| LMArena Japanese | 1209 | 1226 |
| LMArena Korean | 1184 | 1269 |
| LMArena Russian | 1251 | 1296 |
| LMArena Spanish | 1261 | 1360 |
| LMArena French | 1283 | — |
Instruction Following GPT-5 Nano leads
GPT-4: 65.3 (#222), GPT-5 Nano: 75.0 (#79)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| LMArena Instruction Following | 1241 | 1306 |
| IFEval | — | 93.2% |
Long Context GPT-4 leads
GPT-4: 37.7 (#212), GPT-5 Nano: 31.3 (#281)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| LMArena Longer Query | 1244 | 1312 |
| Fiction.LiveBench | — | 44.4% |
Writing & Preference GPT-5 Nano leads
GPT-4: 34.9 (#268), GPT-5 Nano: 39.1 (#249)
| Benchmark | GPT-4 | GPT-5 Nano |
|---|---|---|
| LMArena Text | 1263 | 1320 |
| LMArena Creative Writing | 1244 | 1249 |
| EQ-Bench Creative Writing | 752 | 705 |
| LMArena Multi-Turn | 1257 | 1311 |
| WildBench | — | 80.6% |
Frequently asked questions
Is GPT-4 better than GPT-5 Nano?
GPT-5 Nano is the stronger model overall, scoring 33.5 to 29.1 on the Noometry Index.
Which is cheaper, GPT-4 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; GPT-4 lists at $30 and $60.
Is GPT-4 or GPT-5 Nano better for coding?
GPT-5 Nano scores higher on coding benchmarks: 33.6 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Nano does, with 400K tokens against 8K.
How many benchmarks do GPT-4 and GPT-5 Nano share?
27 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and GPT-5 Nano has 49.