Model comparison
GPT-4.5 vs GPT-5.4
GPT-5.4 is the stronger model overall, scoring 59.5 to 36.9 on the Noometry Index.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. GPT-4.5 scores higher in 0 categories and GPT-5.4 in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 leads 61.7 to 14.1.
- The biggest single-benchmark swing is ARC-AGI-1: 10.3% for GPT-4.5 and 93.7% for GPT-5.4.
Side by side
| GPT-4.5 | GPT-5.4 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 36.9 | 59.5 |
| Released | 2025-02-27 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | — | 1.05M |
| Max output | — | 128K |
| Input $ / M tokens | — | $2.50 |
| Output $ / M tokens | — | $15 |
| Results tracked | 42 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 leads
GPT-4.5: 41.8 (#102), GPT-5.4: 52.8 (#28)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| WeirdML | 39.4% | 77.7% |
| LMArena Coding | 1397 | 1497 |
| SWE-bench Verified | — | 76.9% |
| DeepSWE | — | 51.8% |
| Aider Polyglot | 44.9% | — |
| LMArena WebDev | — | 1465 |
| SciCode | — | 56.6% |
| GSO | — | 31.4% |
| LiveBench Coding | 75.2% | — |
| MirrorCode | — | 15.6% |
| ALE-Bench | — | 1,607 |
| AlgoTune | — | 1.85 |
Agentic & Tool Use GPT-5.4 leads
GPT-4.5: 27.6 (#92), GPT-5.4: 46.5 (#12)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| Terminal-Bench | — | 81.8% |
| APEX-Agents | — | 52.4% |
| τ²-bench Banking | — | 39.4% |
| Cybench | 17.5% | — |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| LMArena Search | — | 1197 |
| METR Time Horizons | — | 74.3% |
| Vending-Bench 2 | — | 6,144 |
Reasoning GPT-5.4 leads
GPT-4.5: 14.1 (#320), GPT-5.4: 61.7 (#17)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 0.8% | 74% |
| ARC-AGI-1 | 10.3% | 93.7% |
| EnigmaEval | 3.2% | 16% |
| LMArena Hard Prompts | 1404 | 1485 |
| Epoch Capabilities Index | 136.74 | 156.92 |
| ForecastBench | 61.7 | 59.5 |
| SimpleBench | 34.5% | — |
| Kagi LLM Benchmark | — | 63.8% |
| NYT Connections (extended) | — | 91.3% |
| CritPt | — | 23.4% |
| Chess Puzzles | — | 44% |
| Thematic Generalization | — | 80% |
| EBR-Bench | — | 25.4% |
| LiveBench Reasoning | 71.1% | — |
| Mystery Game Puzzles | — | 37% |
| DTBench | — | 94.4% |
| LiveBench Data Analysis | 64.3% | — |
| LMCA | — | 52% |
| LiveBench | 69% | — |
Math GPT-5.4 leads
GPT-4.5: 31.0 (#200), GPT-5.4: 73.3 (#14)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 97.8% |
| LMArena Math | 1412 | 1488 |
| FrontierMath (Tiers 1-3) | — | 78.6% |
| FrontierMath Tier 4 | — | 49% |
| MathArena Final-Answer Competitions | — | 83.1% |
| ProofBench | — | 56% |
| LiveBench Math | 69.3% | — |
| MATH Level 5 | 78.6% | — |
| FrontierMath (Feb 2025 set) | — | 47.6% |
| FrontierMath Tier 4 (v1) | — | 27.1% |
Knowledge GPT-5.4 leads
GPT-4.5: 32.4 (#205), GPT-5.4: 65.3 (#13)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 68.7% | 93.3% |
| Humanity's Last Exam | 5.4% | 36.2% |
| LMArena Expert | 1394 | 1507 |
| SimpleQA Verified | — | 45.1% |
| Confabulations | 13.6% | — |
| Vectara Hallucination Rate | — | 7% |
Multimodal GPT-5.4 leads
GPT-4.5: 37.0 (#68), GPT-5.4: 43.8 (#19)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1195 | 1302 |
| VPCT | 45% | — |
| Blueprint-Bench 2 | — | 27.1% |
| Furniture Assembly | — | 37.5% |
| LMArena Document | — | 1471 |
Multilingual GPT-5.4 leads
GPT-4.5: 52.7 (#78), GPT-5.4: 56.4 (#22)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1413 | 1465 |
| LMArena Chinese | 1421 | 1520 |
| LMArena French | 1417 | 1493 |
| LMArena German | 1457 | 1472 |
| LMArena Japanese | 1416 | 1484 |
| LMArena Korean | 1392 | 1449 |
| LMArena Russian | 1419 | 1480 |
| LMArena Spanish | — | 1453 |
Instruction Following GPT-5.4 leads
GPT-4.5: 72.8 (#129), GPT-5.4: 77.3 (#25)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1404 | 1469 |
| LiveBench Instruction Following | 72.3% | — |
Long Context GPT-5.4 leads
GPT-4.5: 40.4 (#150), GPT-5.4: 50.3 (#8)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1406 | 1474 |
| Fiction.LiveBench | 63.9% | — |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |
Writing & Preference GPT-5.4 leads
GPT-4.5: 57.1 (#129), GPT-5.4: 72.1 (#18)
| Benchmark | GPT-4.5 | GPT-5.4 |
|---|---|---|
| LMArena Text | 1417 | 1469 |
| LMArena Creative Writing | 1394 | 1439 |
| EQ-Bench Creative Writing | 1258 | 1840 |
| LMArena Multi-Turn | 1444 | 1482 |
| Short-Story Creative Writing | 75.6% | — |
| EQ-Bench 4 | — | 1272 |
| LiveBench Language | 61.5% | — |
Frequently asked questions
Is GPT-4.5 better than GPT-5.4?
GPT-5.4 is the stronger model overall, scoring 59.5 to 36.9 on the Noometry Index.
Is GPT-4.5 or GPT-5.4 better for coding?
GPT-5.4 scores higher on coding benchmarks: 52.8 versus 41.8 in the Noometry coding category.
How many benchmarks do GPT-4.5 and GPT-5.4 share?
27 benchmarks have published results for both models. GPT-4.5 has 42 scored results on Noometry and GPT-5.4 has 68.