Model comparison
GPT-4 vs GPT-5.5
GPT-5.5 is the stronger model overall, scoring 63.4 to 29.1 on the Noometry Index.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. GPT-4 scores higher in 0 categories and GPT-5.5 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.5 leads 81.7 to 10.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.1% for GPT-4 and 100% for GPT-5.5.
- GPT-5.5 is cheaper at $5 / $30 per million input/output tokens, against $30 / $60 for GPT-4.
- GPT-5.5 accepts more context: 1.05M tokens versus 8K.
Side by side
| GPT-4 | GPT-5.5 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 29.1 | 63.4 |
| Released | 2023-03-14 | 2026-04-23 |
| Weights | Proprietary | Proprietary |
| Context window | 8K | 1.05M |
| Max output | 8K | 128K |
| Input $ / M tokens | $30 | $5 |
| Output $ / M tokens | $60 | $30 |
| Results tracked | 38 | 71 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.5 leads
GPT-4: 31.6 (#283), GPT-5.5: 58.2 (#17)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| WeirdML | 12.4% | 84.9% |
| LMArena Coding | 1254 | 1494 |
| SWE-bench Verified | — | 80.6% |
| DeepSWE | — | 67% |
| FrontierCode | — | 43% |
| LMArena WebDev | — | 1513 |
| SciCode | — | 56.1% |
| GSO | — | 40.2% |
| BigCodeBench Instruct | 46% | — |
| MirrorCode | — | 10% |
| BigCodeBench Complete | 57.2% | — |
| ALE-Bench | — | 1,943 |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, GPT-5.5: 50.7 (#6)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| Terminal-Bench | — | 84.7% |
| APEX-Agents | — | 55.1% |
| OSWorld 2.0 | — | 13% |
| Remote Labor Index | — | 6.3% |
| τ²-bench Banking | — | 44.6% |
| DeepResearch Bench | — | 54% |
| PostTrainBench | — | 27.2% |
| ExploitBench | — | 47.4% |
| GBAEval | — | 53.2% |
| GDP.pdf | — | 26% |
| LMArena Search | — | 1242 |
| METR Time Horizons | 36.1% | — |
| Vending-Bench 2 | — | 7,524 |
Reasoning GPT-5.5 leads
GPT-4: 17.8 (#289), GPT-5.5: 72.8 (#11)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| Chess Puzzles | 4% | 54% |
| LMArena Hard Prompts | 1241 | 1489 |
| Mystery Game Puzzles | 12% | 56% |
| DTBench | 62.7% | 96% |
| LMCA | 17.1% | 54.3% |
| Epoch Capabilities Index | 125.89 | 159.1 |
| ForecastBench | 57.8 | 60.6 |
| ARC-AGI-2 | — | 85% |
| SimpleBench | — | 69% |
| Kagi LLM Benchmark | — | 88.8% |
| NYT Connections (extended) | — | 96.2% |
| ARC-AGI-1 | — | 95% |
| CritPt | — | 27.1% |
| EBR-Bench | — | 34.3% |
| Surface Evolver Bench | — | 88.1% |
| Bench to the Future 3 | — | 0.14 |
| BIG-Bench Hard | 75.1% | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math GPT-5.5 leads
GPT-4: 10.8 (#309), GPT-5.5: 81.7 (#11)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.1% | 100% |
| LMArena Math | 1269 | 1486 |
| FrontierMath (Tiers 1-3) | — | 85.3% |
| FrontierMath Tier 4 | — | 72.5% |
| MathArena Final-Answer Competitions | — | 94.3% |
| ProofBench | — | 50% |
| MATH Level 5 | 23% | — |
| FrontierMath (Feb 2025 set) | — | 51.7% |
| FrontierMath Erdős | — | 0% |
| FrontierMath Tier 4 (v1) | — | 35.4% |
| GSM8K | 92% | — |
Knowledge GPT-5.5 leads
GPT-4: 18.4 (#282), GPT-5.5: 64.4 (#17)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| GPQA Diamond | 35.7% | 94% |
| LMArena Expert | 1211 | 1508 |
| SimpleQA Verified | — | 63% |
| Vectara Hallucination Rate | — | 9.3% |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, GPT-5.5: 46.9 (#12)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| LMArena Vision | — | 1297 |
| Blueprint-Bench 2 | — | 36.2% |
| Furniture Assembly | — | 44.2% |
| LMArena Document | — | 1486 |
Multilingual GPT-5.5 leads
GPT-4: 40.6 (#215), GPT-5.5: 56.4 (#20)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| LMArena Non-English | 1246 | 1467 |
| LMArena Chinese | 1242 | 1533 |
| LMArena French | 1283 | 1486 |
| LMArena German | 1251 | 1480 |
| LMArena Japanese | 1209 | 1498 |
| LMArena Korean | 1184 | 1460 |
| LMArena Russian | 1251 | 1473 |
| LMArena Spanish | 1261 | 1468 |
Instruction Following GPT-5.5 leads
GPT-4: 65.3 (#222), GPT-5.5: 77.5 (#18)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| LMArena Instruction Following | 1241 | 1479 |
Long Context GPT-5.5 leads
GPT-4: 37.7 (#212), GPT-5.5: 48.3 (#12)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| LMArena Longer Query | 1244 | 1484 |
| CL-bench Life | — | 22.2% |
Writing & Preference GPT-5.5 leads
GPT-4: 34.9 (#268), GPT-5.5: 72.7 (#13)
| Benchmark | GPT-4 | GPT-5.5 |
|---|---|---|
| LMArena Text | 1263 | 1472 |
| LMArena Creative Writing | 1244 | 1455 |
| EQ-Bench Creative Writing | 752 | 1844 |
| LMArena Multi-Turn | 1257 | 1476 |
| EQ-Bench 4 | — | 1315 |
Frequently asked questions
Is GPT-4 better than GPT-5.5?
GPT-5.5 is the stronger model overall, scoring 63.4 to 29.1 on the Noometry Index.
Which is cheaper, GPT-4 or GPT-5.5?
GPT-5.5 is cheaper. It lists at $5 per million input tokens and $30 per million output tokens; GPT-4 lists at $30 and $60.
Is GPT-4 or GPT-5.5 better for coding?
GPT-5.5 scores higher on coding benchmarks: 58.2 versus 31.6 in the Noometry coding category.
Which has the bigger context window?
GPT-5.5 does, with 1.05M tokens against 8K.
How many benchmarks do GPT-4 and GPT-5.5 share?
27 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and GPT-5.5 has 71.