Model comparison
GPT-5 Nano vs o3-mini
o3-mini is the stronger model overall, scoring 36.7 to 33.4 on the Noometry Index. GPT-5 Nano costs 14× less per token, which makes it the better buy when o3-mini's lead doesn't matter for your workload.
Last verified . 33 shared benchmarks.
Summary
- They share 33 benchmarks with published results for both. GPT-5 Nano scores higher in 2 categories and o3-mini in 7 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where o3-mini leads 50.4 to 39.4.
- The biggest single-benchmark swing is ARC-AGI-1: 20.7% for GPT-5 Nano and 34.5% for o3-mini.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $1.10 / $4.40 for o3-mini.
- GPT-5 Nano accepts more context: 400K tokens versus 200K.
Side by side
| GPT-5 Nano | o3-mini | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 33.4 | 36.7 |
| Released | 2025-08-07 | 2024-12-20 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 200K |
| Max output | 128K | 100K |
| Input $ / M tokens | $0.05 | $1.10 |
| Output $ / M tokens | $0.40 | $4.40 |
| Results tracked | 48 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3-mini leads
GPT-5 Nano: 33.3 (#248), o3-mini: 41.4 (#116)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| WeirdML | 38.1% | 43.7% |
| LMArena Coding | 1351 | 1378 |
| SWE-bench Verified (bash only) | 34.8% | — |
| Aider Polyglot | — | 60.4% |
| SciCode | — | 39.8% |
| GSO | — | 1.3% |
| LiveBench Coding | — | 82.7% |
| CadEval | — | 54% |
| ALE-Bench | 718.67 | — |
Agentic & Tool Use o3-mini leads
GPT-5 Nano: 25.6 (#99), o3-mini: 29.3 (#79)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| Terminal-Bench | 21.8% | — |
| Berkeley Function Calling Leaderboard | 51.5% | — |
| Cybench | — | 22.5% |
Reasoning Too close to call
GPT-5 Nano: 16.4 (#291), o3-mini: 16.3 (#295)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| ARC-AGI-2 | 2.6% | 3% |
| ARC-AGI-1 | 20.7% | 34.5% |
| Chess Puzzles | 27% | 17% |
| LMArena Hard Prompts | 1329 | 1366 |
| Mystery Game Puzzles | 9% | 7% |
| DTBench | 62.7% | 68.8% |
| LMCA | 7.9% | 19% |
| Epoch Capabilities Index | 139.39 | 140.35 |
| ForecastBench | 59.1 | 59.6 |
| SimpleBench | — | 22.8% |
| Kagi LLM Benchmark | 62.2% | — |
| CritPt | — | 0.3% |
| LiveBench Reasoning | — | 89.6% |
| LiveBench Data Analysis | — | 70.6% |
| LiveBench | — | 75.9% |
Math GPT-5 Nano leads
GPT-5 Nano: 29.1 (#221), o3-mini: 27.6 (#236)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 20% | 18.6% |
| FrontierMath Tier 4 | 2.4% | 0% |
| OTIS Mock AIME 2024-2025 | 81.1% | 76.9% |
| LMArena Math | 1318 | 1396 |
| MATH Level 5 | 95.2% | 96.5% |
| FrontierMath (Feb 2025 set) | 8.3% | 12.4% |
| FrontierMath Tier 4 (v1) | 2.1% | 4.2% |
| ProofBench | 12% | — |
| Omni-MATH | 54.6% | — |
| LiveBench Math | — | 77.3% |
Knowledge o3-mini leads
GPT-5 Nano: 35.8 (#166), o3-mini: 38.3 (#135)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| GPQA Diamond | 69.4% | 77% |
| SimpleQA Verified | 11.7% | 15.3% |
| LMArena Expert | 1321 | 1364 |
| MMLU-Pro | 77.8% | — |
| Confabulations | — | 17.9% |
| Vectara Hallucination Rate | 10.5% | — |
| GPQA (HELM) | 67.9% | — |
Multimodal Not comparable
GPT-5 Nano: 30.7 (#102), o3-mini: —
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| LMArena Vision | 1159 | — |
| VPCT | 37.2% | — |
Multilingual Too close to call
GPT-5 Nano: 45.5 (#167), o3-mini: 45.9 (#159)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| LMArena Non-English | 1313 | 1319 |
| LMArena Chinese | 1355 | 1379 |
| LMArena German | 1329 | 1303 |
| LMArena Korean | 1269 | 1314 |
| LMArena Russian | 1296 | 1304 |
| LMArena Spanish | 1360 | 1321 |
| LMArena French | — | 1333 |
| LMArena Japanese | — | 1286 |
Instruction Following Too close to call
GPT-5 Nano: 75.2 (#72), o3-mini: 75.3 (#67)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| LMArena Instruction Following | 1306 | 1338 |
| LiveBench Instruction Following | — | 84.4% |
| IFEval | 93.2% | — |
Long Context o3-mini leads
GPT-5 Nano: 31.3 (#275), o3-mini: 33.8 (#251)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| Fiction.LiveBench | 44.4% | 50% |
| LMArena Longer Query | 1313 | 1343 |
Writing & Preference o3-mini leads
GPT-5 Nano: 39.4 (#244), o3-mini: 50.4 (#177)
| Benchmark | GPT-5 Nano | o3-mini |
|---|---|---|
| LMArena Text | 1320 | 1337 |
| LMArena Creative Writing | 1250 | 1286 |
| LMArena Multi-Turn | 1311 | 1320 |
| Short-Story Creative Writing | — | 61.7% |
| EQ-Bench Creative Writing | 705 | — |
| WildBench | 80.6% | — |
| LiveBench Language | — | 50.7% |
Frequently asked questions
Is GPT-5 Nano better than o3-mini?
o3-mini is the stronger model overall, scoring 36.7 to 33.4 on the Noometry Index. GPT-5 Nano costs 14× less per token, which makes it the better buy when o3-mini's lead doesn't matter for your workload.
Which is cheaper, GPT-5 Nano or o3-mini?
GPT-5 Nano is cheaper. It lists at $0.05 per million input tokens and $0.40 per million output tokens; o3-mini lists at $1.10 and $4.40.
Is GPT-5 Nano or o3-mini better for coding?
o3-mini scores higher on coding benchmarks: 41.4 versus 33.3 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-mini share?
33 benchmarks have published results for both models. GPT-5 Nano has 48 scored results on Noometry and o3-mini has 51.