Model comparison
GPT-5 vs o4-mini
GPT-5 is the stronger model overall, scoring 50.9 to 41.6 on the Noometry Index. o4-mini costs 1.8× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 58 shared benchmarks.
Summary
- They share 58 benchmarks with published results for both. GPT-5 scores higher in 9 categories and o4-mini in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 45.5.
- The biggest single-benchmark swing is SimpleQA Verified: 50.1% for GPT-5 and 19.6% for o4-mini.
- o4-mini is cheaper at $1.10 / $4.40 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 200K.
Side by side
| GPT-5 | o4-mini | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 50.9 | 41.6 |
| Released | 2025-08-07 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 200K |
| Max output | 128K | 100K |
| Input $ / M tokens | $1.25 | $1.10 |
| Output $ / M tokens | $10 | $4.40 |
| Results tracked | 69 | 60 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
GPT-5: 50.3 (#47), o4-mini: 40.9 (#127)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| SWE-bench Verified (bash only) | 65% | 45% |
| Aider Polyglot | 88% | 72% |
| GSO | 6.9% | 3.6% |
| WeirdML | 60.7% | 52.6% |
| LMArena Coding | 1436 | 1368 |
| ALE-Bench | 1,162 | 826.17 |
| AlgoTune | 1.67 | 1.72 |
| SWE-bench Verified | 73.6% | — |
| LMArena WebDev | 1418 | — |
| SciCode | 42.9% | — |
| CadEval | — | 62% |
Agentic & Tool Use Too close to call
GPT-5: 33.1 (#56), o4-mini: 32.6 (#61)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| GDPval | 34.8% | 25.3% |
| METR Time Horizons | 69.6% | 63.9% |
| Terminal-Bench | 49.6% | — |
| Berkeley Function Calling Leaderboard | — | 53.2% |
| Remote Labor Index | 1.7% | — |
| DeepResearch Bench | 49.6% | — |
| BALROG | 32.8% | — |
| LMArena Search | 1133 | — |
Reasoning GPT-5 leads
GPT-5: 38.3 (#64), o4-mini: 24.6 (#162)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| ARC-AGI-2 | 9.9% | 6.1% |
| SimpleBench | 56.7% | 38.7% |
| Kagi LLM Benchmark | 72.7% | 67.6% |
| ARC-AGI-1 | 65.7% | 58.7% |
| CritPt | 12.6% | 0.6% |
| Chess Puzzles | 37% | 26% |
| EnigmaEval | 10.5% | 9.2% |
| LMArena Hard Prompts | 1416 | 1351 |
| Mystery Game Puzzles | 23% | 5% |
| DTBench | 90.7% | 77.6% |
| LMCA | 40% | 26.5% |
| Epoch Capabilities Index | 150 | 145.64 |
| ForecastBench | 61.4 | 61.8 |
| EBR-Bench | 12.7% | — |
Math GPT-5 leads
GPT-5: 55.0 (#44), o4-mini: 40.8 (#89)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| FrontierMath (Tiers 1-3) | 55.4% | 36.1% |
| FrontierMath Tier 4 | 22% | 4.9% |
| OTIS Mock AIME 2024-2025 | 91.4% | 81.7% |
| Omni-MATH | 64.7% | 72% |
| LMArena Math | 1407 | 1389 |
| MATH Level 5 | 98.1% | 97.8% |
| FrontierMath (Feb 2025 set) | 32.4% | 24.8% |
| FrontierMath Tier 4 (v1) | 12.5% | 6.3% |
| ProofBench | 18% | — |
Knowledge GPT-5 leads
GPT-5: 56.6 (#43), o4-mini: 43.6 (#91)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| GPQA Diamond | 86.2% | 79.6% |
| Humanity's Last Exam | 25.3% | 18.1% |
| SimpleQA Verified | 50.1% | 19.6% |
| MMLU-Pro | 86.3% | 82% |
| Confabulations | 10.3% | 15.8% |
| Vectara Hallucination Rate | 14.7% | 18.6% |
| GPQA (HELM) | 79.2% | 73.5% |
| LMArena Expert | 1419 | 1343 |
Multimodal GPT-5 leads
GPT-5: 46.8 (#13), o4-mini: 40.2 (#49)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| LMArena Vision | 1232 | 1194 |
| GeoBench | 81% | 64% |
| VPCT | 66% | 57.5% |
Multilingual GPT-5 leads
GPT-5: 51.4 (#110), o4-mini: 47.0 (#154)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| LMArena Non-English | 1397 | 1337 |
| LMArena Chinese | 1422 | 1354 |
| LMArena French | 1410 | 1364 |
| LMArena German | 1416 | 1336 |
| LMArena Japanese | 1409 | 1308 |
| LMArena Korean | 1360 | 1312 |
| LMArena Russian | 1406 | 1334 |
| LMArena Spanish | 1399 | 1347 |
Instruction Following o4-mini leads
GPT-5: 73.8 (#113), o4-mini: 75.2 (#68)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| IFEval | 87.5% | 92.8% |
| LMArena Instruction Following | 1388 | 1321 |
Long Context GPT-5 leads
GPT-5: 69.5 (#2), o4-mini: 45.5 (#33)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| Fiction.LiveBench | 97.2% | 77.8% |
| LMArena Longer Query | 1399 | 1315 |
Writing & Preference GPT-5 leads
GPT-5: 63.4 (#65), o4-mini: 54.0 (#152)
| Benchmark | GPT-5 | o4-mini |
|---|---|---|
| LMArena Text | 1406 | 1353 |
| LMArena Creative Writing | 1365 | 1294 |
| Short-Story Creative Writing | 86% | 75% |
| WildBench | 85.7% | 85.4% |
| LMArena Multi-Turn | 1426 | 1350 |
| EQ-Bench Creative Writing | 1627 | — |
Frequently asked questions
Is GPT-5 better than o4-mini?
GPT-5 is the stronger model overall, scoring 50.9 to 41.6 on the Noometry Index. o4-mini costs 1.8× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, GPT-5 or o4-mini?
o4-mini is cheaper. It lists at $1.10 per million input tokens and $4.40 per million output tokens; GPT-5 lists at $1.25 and $10.
Is GPT-5 or o4-mini better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 40.9 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 200K.
How many benchmarks do GPT-5 and o4-mini share?
58 benchmarks have published results for both models. GPT-5 has 69 scored results on Noometry and o4-mini has 60.