Model comparison
GPT-4 Turbo vs GPT-5
GPT-5 is the stronger model overall, scoring 50.9 to 30.3 on the Noometry Index.
Last verified . 31 shared benchmarks.
Summary
- They share 31 benchmarks with published results for both. GPT-4 Turbo scores higher in 0 categories and GPT-5 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5 leads 54.6 to 8.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 6.7% for GPT-4 Turbo and 91.4% for GPT-5.
- GPT-5 is cheaper at $1.25 / $10 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- GPT-5 accepts more context: 400K tokens versus 128K.
Side by side
| GPT-4 Turbo | GPT-5 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 30.3 | 50.9 |
| Released | 2023-11-06 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 128K | 400K |
| Max output | 4K | 128K |
| Input $ / M tokens | $10 | $1.25 |
| Output $ / M tokens | $30 | $10 |
| Results tracked | 36 | 69 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
GPT-4 Turbo: 33.4 (#247), GPT-5: 50.7 (#39)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| WeirdML | 18% | 60.7% |
| LMArena Coding | 1268 | 1436 |
| SWE-bench Verified | — | 73.6% |
| SWE-bench Verified (bash only) | — | 65% |
| Aider Polyglot | — | 88% |
| LMArena WebDev | — | 1417 |
| SciCode | — | 42.9% |
| GSO | — | 6.9% |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 58.2% | — |
| ALE-Bench | — | 1,162 |
| AlgoTune | — | 1.67 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73.3% | — |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, GPT-5: 33.0 (#53)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| METR Time Horizons | 36.7% | 69.6% |
| Terminal-Bench | — | 49.6% |
| GDPval | — | 34.8% |
| Remote Labor Index | — | 1.7% |
| DeepResearch Bench | — | 49.6% |
| BALROG | — | 32.8% |
| LMArena Search | — | 1133 |
Reasoning GPT-5 leads
GPT-4 Turbo: 14.9 (#311), GPT-5: 38.3 (#55)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| SimpleBench | 25.1% | 56.7% |
| Chess Puzzles | 6% | 37% |
| LMArena Hard Prompts | 1251 | 1415 |
| DTBench | 61.6% | 90.7% |
| LMCA | 9.8% | 40% |
| Epoch Capabilities Index | 127.25 | 150 |
| ForecastBench | 59.4 | 61.4 |
| ARC-AGI-2 | — | 9.9% |
| Kagi LLM Benchmark | — | 72.7% |
| ARC-AGI-1 | — | 65.7% |
| CritPt | — | 12.6% |
| EnigmaEval | — | 10.5% |
| EBR-Bench | — | 12.7% |
| Mystery Game Puzzles | — | 23% |
Math GPT-5 leads
GPT-4 Turbo: 8.6 (#316), GPT-5: 54.6 (#36)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 0.7% | 55.4% |
| OTIS Mock AIME 2024-2025 | 6.7% | 91.4% |
| LMArena Math | 1272 | 1406 |
| MATH Level 5 | 46.7% | 98.1% |
| FrontierMath Tier 4 | — | 22% |
| ProofBench | — | 18% |
| Omni-MATH | — | 64.7% |
| FrontierMath (Feb 2025 set) | — | 32.4% |
| FrontierMath Tier 4 (v1) | — | 12.5% |
Knowledge GPT-5 leads
GPT-4 Turbo: 24.3 (#263), GPT-5: 56.5 (#37)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| GPQA Diamond | 46.6% | 86.2% |
| Confabulations | 28.4% | 10.3% |
| LMArena Expert | 1223 | 1420 |
| Humanity's Last Exam | — | 25.3% |
| SimpleQA Verified | — | 50.1% |
| MMLU-Pro | — | 86.3% |
| Vectara Hallucination Rate | — | 14.7% |
| GPQA (HELM) | — | 79.2% |
| MMLU | 81.3% | — |
Multimodal GPT-5 leads
GPT-4 Turbo: 30.0 (#104), GPT-5: 46.1 (#12)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| LMArena Vision | 1090 | 1232 |
| GeoBench | — | 81% |
| VPCT | — | 66% |
Multilingual GPT-5 leads
GPT-4 Turbo: 40.7 (#211), GPT-5: 51.5 (#105)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| LMArena Non-English | 1245 | 1397 |
| LMArena Chinese | 1242 | 1423 |
| LMArena French | 1275 | 1409 |
| LMArena German | 1259 | 1415 |
| LMArena Japanese | 1194 | 1411 |
| LMArena Korean | 1187 | 1361 |
| LMArena Russian | 1259 | 1406 |
| LMArena Spanish | 1260 | 1398 |
Instruction Following GPT-5 leads
GPT-4 Turbo: 66.0 (#211), GPT-5: 74.0 (#107)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| LMArena Instruction Following | 1250 | 1388 |
| IFEval | — | 87.5% |
Long Context GPT-5 leads
GPT-4 Turbo: 38.0 (#200), GPT-5: 69.5 (#2)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| LMArena Longer Query | 1254 | 1399 |
| Fiction.LiveBench | — | 97.2% |
Writing & Preference GPT-5 leads
GPT-4 Turbo: 47.9 (#201), GPT-5: 63.5 (#63)
| Benchmark | GPT-4 Turbo | GPT-5 |
|---|---|---|
| LMArena Text | 1272 | 1406 |
| LMArena Creative Writing | 1269 | 1365 |
| LMArena Multi-Turn | 1267 | 1426 |
| Short-Story Creative Writing | — | 86% |
| EQ-Bench Creative Writing | — | 1627 |
| WildBench | — | 85.7% |
Frequently asked questions
Is GPT-4 Turbo better than GPT-5?
GPT-5 is the stronger model overall, scoring 50.9 to 30.3 on the Noometry Index.
Which is cheaper, GPT-4 Turbo or GPT-5?
GPT-5 is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; GPT-4 Turbo lists at $10 and $30.
Is GPT-4 Turbo or GPT-5 better for coding?
GPT-5 scores higher on coding benchmarks: 50.7 versus 33.4 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 128K.
How many benchmarks do GPT-4 Turbo and GPT-5 share?
31 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and GPT-5 has 69.