Model comparison
GPT-4.1 vs GPT-5
GPT-5 is the stronger model overall, scoring 50.9 to 35.9 on the Noometry Index.
Last verified . 50 shared benchmarks.
Summary
- They share 50 benchmarks with published results for both. GPT-4.1 scores higher in 1 category and GPT-5 in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5 leads 55.0 to 22.3.
- The biggest single-benchmark swing is ARC-AGI-1: 5.5% for GPT-4.1 and 65.7% for GPT-5.
- Both cost about the same: $2 input and $8 output per million tokens.
- GPT-4.1 accepts more context: 1.05M tokens versus 400K.
Side by side
| GPT-4.1 | GPT-5 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 35.9 | 50.9 |
| Released | 2025-04-14 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 400K |
| Max output | 33K | 128K |
| Input $ / M tokens | $2 | $1.25 |
| Output $ / M tokens | $8 | $10 |
| Results tracked | 52 | 69 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
GPT-4.1: 34.4 (#238), GPT-5: 50.3 (#47)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| SWE-bench Verified | 48.5% | 73.6% |
| SWE-bench Verified (bash only) | 39.6% | 65% |
| Aider Polyglot | 52.4% | 88% |
| WeirdML | 39% | 60.7% |
| LMArena Coding | 1391 | 1436 |
| ALE-Bench | 558.1 | 1,162 |
| LMArena WebDev | — | 1418 |
| SciCode | — | 42.9% |
| GSO | — | 6.9% |
| CadEval | 42% | — |
| AlgoTune | — | 1.67 |
Agentic & Tool Use GPT-4.1 leads
GPT-4.1: 34.7 (#43), GPT-5: 33.1 (#56)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| Terminal-Bench | — | 49.6% |
| Berkeley Function Calling Leaderboard | 54% | — |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| DeepResearch Bench | — | 49.6% |
| BALROG | — | 32.8% |
| LMArena Search | — | 1133 |
| METR Time Horizons | — | 69.6% |
Reasoning GPT-5 leads
GPT-4.1: 11.7 (#339), GPT-5: 38.3 (#64)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| ARC-AGI-2 | 0.4% | 9.9% |
| SimpleBench | 27% | 56.7% |
| Kagi LLM Benchmark | 52.3% | 72.7% |
| ARC-AGI-1 | 5.5% | 65.7% |
| Chess Puzzles | 6% | 37% |
| EnigmaEval | 2.2% | 10.5% |
| LMArena Hard Prompts | 1384 | 1416 |
| DTBench | 68.3% | 90.7% |
| LMCA | 25.6% | 40% |
| Epoch Capabilities Index | 136.78 | 150 |
| ForecastBench | 61.5 | 61.4 |
| CritPt | — | 12.6% |
| EBR-Bench | — | 12.7% |
| Mystery Game Puzzles | — | 23% |
Math GPT-5 leads
GPT-4.1: 22.3 (#280), GPT-5: 55.0 (#44)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 6% | 55.4% |
| OTIS Mock AIME 2024-2025 | 38.3% | 91.4% |
| Omni-MATH | 47.1% | 64.7% |
| LMArena Math | 1370 | 1407 |
| MATH Level 5 | 83% | 98.1% |
| FrontierMath (Feb 2025 set) | 5.5% | 32.4% |
| FrontierMath Tier 4 (v1) | 0% | 12.5% |
| FrontierMath Tier 4 | — | 22% |
| ProofBench | — | 18% |
Knowledge GPT-5 leads
GPT-4.1: 37.1 (#160), GPT-5: 56.6 (#43)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| GPQA Diamond | 66.9% | 86.2% |
| Humanity's Last Exam | 5.4% | 25.3% |
| SimpleQA Verified | 31.1% | 50.1% |
| MMLU-Pro | 81.1% | 86.3% |
| Vectara Hallucination Rate | 5.6% | 14.7% |
| GPQA (HELM) | 65.9% | 79.2% |
| LMArena Expert | 1364 | 1419 |
| Confabulations | — | 10.3% |
Multimodal GPT-5 leads
GPT-4.1: 38.2 (#67), GPT-5: 46.8 (#13)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| LMArena Vision | 1211 | 1232 |
| GeoBench | 72% | 81% |
| VPCT | — | 66% |
Multilingual GPT-5 leads
GPT-4.1: 49.4 (#133), GPT-5: 51.4 (#110)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| LMArena Non-English | 1370 | 1397 |
| LMArena Chinese | 1382 | 1422 |
| LMArena French | 1382 | 1410 |
| LMArena German | 1381 | 1416 |
| LMArena Japanese | 1319 | 1409 |
| LMArena Korean | 1339 | 1360 |
| LMArena Russian | 1377 | 1406 |
| LMArena Spanish | 1376 | 1399 |
Instruction Following GPT-5 leads
GPT-4.1: 71.3 (#153), GPT-5: 73.8 (#113)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| IFEval | 83.8% | 87.5% |
| LMArena Instruction Following | 1367 | 1388 |
Long Context GPT-5 leads
GPT-4.1: 40.0 (#163), GPT-5: 69.5 (#2)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| Fiction.LiveBench | 63.9% | 97.2% |
| LMArena Longer Query | 1385 | 1399 |
Writing & Preference GPT-5 leads
GPT-4.1: 57.6 (#125), GPT-5: 63.4 (#65)
| Benchmark | GPT-4.1 | GPT-5 |
|---|---|---|
| LMArena Text | 1383 | 1406 |
| LMArena Creative Writing | 1363 | 1365 |
| EQ-Bench Creative Writing | 1420 | 1627 |
| WildBench | 85.4% | 85.7% |
| LMArena Multi-Turn | 1398 | 1426 |
| Short-Story Creative Writing | — | 86% |
Frequently asked questions
Is GPT-4.1 better than GPT-5?
GPT-5 is the stronger model overall, scoring 50.9 to 35.9 on the Noometry Index.
Which is cheaper, GPT-4.1 or GPT-5?
GPT-5 is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; GPT-4.1 lists at $2 and $8.
Is GPT-4.1 or GPT-5 better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 34.4 in the Noometry coding category.
Which has the bigger context window?
GPT-4.1 does, with 1.05M tokens against 400K.
How many benchmarks do GPT-4.1 and GPT-5 share?
50 benchmarks have published results for both models. GPT-4.1 has 52 scored results on Noometry and GPT-5 has 69.