Model comparison
GPT-5 Nano vs o3
o3 is the stronger model overall, scoring 47.5 to 33.5 on the Noometry Index. GPT-5 Nano costs 25× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Last verified . 45 shared benchmarks.
Summary
- They share 45 benchmarks with published results for both. GPT-5 Nano scores higher in 1 category and o3 in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where o3 leads 63.5 to 39.1.
- The biggest single-benchmark swing is Fiction.LiveBench: 44.4% for GPT-5 Nano and 88.9% for o3.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $2 / $8 for o3.
- GPT-5 Nano accepts more context: 400K tokens versus 200K.
Side by side
| GPT-5 Nano | o3 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 33.5 | 47.5 |
| Released | 2025-08-07 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 200K |
| Max output | 128K | 100K |
| Input $ / M tokens | $0.05 | $2 |
| Output $ / M tokens | $0.40 | $8 |
| Results tracked | 49 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
GPT-5 Nano: 33.6 (#254), o3: 46.8 (#64)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| SWE-bench Verified (bash only) | 34.8% | 58.4% |
| WeirdML | 38.1% | 52.4% |
| LMArena Coding | 1351 | 1408 |
| ALE-Bench | 718.67 | 933.55 |
| SWE-bench Verified | — | 62.3% |
| Aider Polyglot | — | 81.3% |
| GSO | — | 8.8% |
| CadEval | — | 74% |
Agentic & Tool Use o3 leads
GPT-5 Nano: 25.8 (#106), o3: 34.5 (#44)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 51.5% | 63% |
| Terminal-Bench | 21.8% | — |
| GDPval | — | 30.8% |
| DeepResearch Bench | — | 45.2% |
| OSWorld | — | 23% |
| LMArena Search | — | 1144 |
| METR Time Horizons | — | 65.4% |
Reasoning o3 leads
GPT-5 Nano: 16.3 (#306), o3: 32.0 (#78)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| ARC-AGI-2 | 2.6% | 6.5% |
| Kagi LLM Benchmark | 62.2% | 67.6% |
| ARC-AGI-1 | 20.7% | 60.8% |
| Chess Puzzles | 27% | 38% |
| LMArena Hard Prompts | 1328 | 1402 |
| Mystery Game Puzzles | 9% | 29% |
| DTBench | 62.7% | 84.8% |
| LMCA | 7.9% | 39.7% |
| Epoch Capabilities Index | 139.38 | 146.86 |
| ForecastBench | 59.1 | 62.5 |
| SimpleBench | — | 53.1% |
| CritPt | — | 1.4% |
| EnigmaEval | — | 13.1% |
Math o3 leads
GPT-5 Nano: 29.4 (#241), o3: 50.2 (#58)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 20% | 33.3% |
| OTIS Mock AIME 2024-2025 | 81.1% | 84.4% |
| Omni-MATH | 54.6% | 71.4% |
| LMArena Math | 1317 | 1426 |
| MATH Level 5 | 95.2% | 97.8% |
| FrontierMath (Feb 2025 set) | 8.3% | 18.7% |
| FrontierMath Tier 4 (v1) | 2.1% | 2.1% |
| FrontierMath Tier 4 | 2.4% | — |
| ProofBench | 12% | — |
Knowledge o3 leads
GPT-5 Nano: 35.9 (#178), o3: 54.6 (#52)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| GPQA Diamond | 69.4% | 81.8% |
| SimpleQA Verified | 11.7% | 49.4% |
| MMLU-Pro | 77.8% | 85.9% |
| GPQA (HELM) | 67.9% | 75.3% |
| LMArena Expert | 1321 | 1402 |
| Humanity's Last Exam | — | 20.3% |
| Confabulations | — | 14.4% |
| Vectara Hallucination Rate | 10.5% | — |
Multimodal o3 leads
GPT-5 Nano: 31.3 (#108), o3: 41.4 (#36)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| LMArena Vision | 1159 | 1214 |
| VPCT | 37.2% | 52% |
| GeoBench | — | 74% |
Multilingual o3 leads
GPT-5 Nano: 45.3 (#172), o3: 51.7 (#105)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| LMArena Non-English | 1313 | 1401 |
| LMArena Chinese | 1356 | 1437 |
| LMArena German | 1327 | 1420 |
| LMArena Japanese | 1226 | 1403 |
| LMArena Korean | 1269 | 1370 |
| LMArena Russian | 1296 | 1406 |
| LMArena Spanish | 1360 | 1395 |
| LMArena French | — | 1430 |
Instruction Following GPT-5 Nano leads
GPT-5 Nano: 75.0 (#79), o3: 72.8 (#127)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| IFEval | 93.2% | 86.9% |
| LMArena Instruction Following | 1306 | 1368 |
Long Context o3 leads
GPT-5 Nano: 31.3 (#281), o3: 53.3 (#6)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| Fiction.LiveBench | 44.4% | 88.9% |
| LMArena Longer Query | 1312 | 1372 |
| CL-bench | — | 17.8% |
Writing & Preference o3 leads
GPT-5 Nano: 39.1 (#249), o3: 63.5 (#64)
| Benchmark | GPT-5 Nano | o3 |
|---|---|---|
| LMArena Text | 1320 | 1410 |
| LMArena Creative Writing | 1249 | 1359 |
| EQ-Bench Creative Writing | 705 | 1676 |
| WildBench | 80.6% | 86.1% |
| LMArena Multi-Turn | 1311 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
Frequently asked questions
Is GPT-5 Nano better than o3?
o3 is the stronger model overall, scoring 47.5 to 33.5 on the Noometry Index. GPT-5 Nano costs 25× less per token, which makes it the better buy when o3's lead doesn't matter for your workload.
Which is cheaper, GPT-5 Nano or o3?
GPT-5 Nano is cheaper. It lists at $0.05 per million input tokens and $0.40 per million output tokens; o3 lists at $2 and $8.
Is GPT-5 Nano or o3 better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 33.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Nano does, with 400K tokens against 200K.
How many benchmarks do GPT-5 Nano and o3 share?
45 benchmarks have published results for both models. GPT-5 Nano has 49 scored results on Noometry and o3 has 63.