Model comparison
Claude Sonnet 4.5 vs GPT-5.4
GPT-5.4 is the stronger model overall, scoring 59.4 to 44.1 on the Noometry Index.
Last verified . 55 shared benchmarks.
Summary
- They share 55 benchmarks with published results for both. Claude Sonnet 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 math, where GPT-5.4 leads 73.5 to 32.3.
- The biggest single-benchmark swing is ARC-AGI-2: 13.6% for Claude Sonnet 4.5 and 74% for GPT-5.4.
- GPT-5.4 is cheaper at $2.50 / $15 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.5.
- GPT-5.4 accepts more context: 1.05M tokens versus 200K.
Side by side
| Claude Sonnet 4.5 | GPT-5.4 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 44.1 | 59.4 |
| Released | 2025-09-29 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1.05M |
| Max output | 64K | 128K |
| Input $ / M tokens | $3 | $2.50 |
| Output $ / M tokens | $15 | $15 |
| Results tracked | 73 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 leads
Claude Sonnet 4.5: 47.3 (#61), GPT-5.4: 52.6 (#33)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| SWE-bench Verified | 71.3% | 76.9% |
| LMArena WebDev | 1393 | 1465 |
| SciCode | 44.7% | 56.6% |
| GSO | 14.7% | 31.4% |
| WeirdML | 47.7% | 77.7% |
| LMArena Coding | 1489 | 1497 |
| ALE-Bench | 796.15 | 1,607 |
| AlgoTune | 1.52 | 1.85 |
| DeepSWE | — | 51.8% |
| SWE-bench Verified (bash only) | 71.4% | — |
| SWE-bench Multilingual | 67% | — |
| MirrorCode | — | 15.6% |
Agentic & Tool Use GPT-5.4 leads
Claude Sonnet 4.5: 38.3 (#32), GPT-5.4: 46.5 (#13)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| Terminal-Bench | 46.5% | 81.8% |
| τ²-bench Banking | 25.3% | 39.4% |
| DeepResearch Bench | 52.6% | 35.1% |
| LMArena Search | 1159 | 1197 |
| METR Time Horizons | 67.4% | 74.3% |
| Vending-Bench 2 | 3,839 | 6,144 |
| APEX-Agents | — | 52.4% |
| Berkeley Function Calling Leaderboard | 73.2% | — |
| GDPval | 42.5% | — |
| Remote Labor Index | 2.1% | — |
| τ²-bench Airline | 72% | — |
| τ²-bench Retail | 72.4% | — |
| τ²-bench Telecom | 84.9% | — |
| Cybench | 60% | — |
| OSWorld | 62.9% | — |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
Reasoning GPT-5.4 leads
Claude Sonnet 4.5: 26.9 (#125), GPT-5.4: 61.8 (#19)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 13.6% | 74% |
| Kagi LLM Benchmark | 57.9% | 63.8% |
| NYT Connections (extended) | 37.3% | 91.3% |
| ARC-AGI-1 | 63.7% | 93.7% |
| CritPt | 1.1% | 23.4% |
| Chess Puzzles | 12% | 44% |
| EnigmaEval | 6% | 16% |
| EBR-Bench | 2.4% | 25.4% |
| LMArena Hard Prompts | 1462 | 1485 |
| Mystery Game Puzzles | 17% | 37% |
| DTBench | 83.2% | 94.4% |
| LMCA | 38.8% | 52% |
| Epoch Capabilities Index | 146.84 | 156.81 |
| ForecastBench | 61.9 | 59.5 |
| SimpleBench | 54.3% | — |
| Thematic Generalization | — | 80% |
Math GPT-5.4 leads
Claude Sonnet 4.5: 32.3 (#216), GPT-5.4: 73.5 (#19)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 23.9% | 78.6% |
| FrontierMath Tier 4 | 2.4% | 49% |
| OTIS Mock AIME 2024-2025 | 77.8% | 97.8% |
| ProofBench | 19% | 56% |
| LMArena Math | 1449 | 1488 |
| FrontierMath (Feb 2025 set) | 15.2% | 47.6% |
| FrontierMath Tier 4 (v1) | 4.2% | 27.1% |
| MathArena Final-Answer Competitions | — | 83.1% |
| Omni-MATH | 55.3% | — |
| MATH Level 5 | 97.7% | — |
Knowledge GPT-5.4 leads
Claude Sonnet 4.5: 48.4 (#76), GPT-5.4: 65.3 (#14)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 82.3% | 93.3% |
| Humanity's Last Exam | 13.7% | 36.2% |
| SimpleQA Verified | 30.7% | 45.1% |
| Vectara Hallucination Rate | 12% | 7% |
| LMArena Expert | 1482 | 1507 |
| MMLU-Pro | 86.9% | — |
| GPQA (HELM) | 68.6% | — |
Multimodal GPT-5.4 leads
Claude Sonnet 4.5: 34.8 (#89), GPT-5.4: 43.7 (#20)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| LMArena Document | 1450 | 1471 |
| LMArena Vision | — | 1303 |
| VPCT | 39.8% | — |
| Blueprint-Bench 2 | — | 27.1% |
| Furniture Assembly | — | 37.5% |
Multilingual GPT-5.4 leads
Claude Sonnet 4.5: 53.4 (#69), GPT-5.4: 56.2 (#23)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1425 | 1465 |
| LMArena Chinese | 1459 | 1519 |
| LMArena French | 1458 | 1493 |
| LMArena German | 1427 | 1472 |
| LMArena Japanese | 1390 | 1485 |
| LMArena Korean | 1403 | 1448 |
| LMArena Russian | 1437 | 1480 |
| LMArena Spanish | 1457 | 1454 |
Instruction Following GPT-5.4 leads
Claude Sonnet 4.5: 75.0 (#78), GPT-5.4: 77.1 (#27)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1459 | 1469 |
| IFEval | 85% | — |
Long Context GPT-5.4 leads
Claude Sonnet 4.5: 45.2 (#46), GPT-5.4: 50.3 (#8)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1476 | 1473 |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |
Writing & Preference GPT-5.4 leads
Claude Sonnet 4.5: 66.5 (#34), GPT-5.4: 71.9 (#17)
| Benchmark | Claude Sonnet 4.5 | GPT-5.4 |
|---|---|---|
| LMArena Text | 1439 | 1469 |
| LMArena Creative Writing | 1442 | 1439 |
| EQ-Bench Creative Writing | 1678 | 1840 |
| LMArena Multi-Turn | 1465 | 1482 |
| WildBench | 85.4% | — |
| EQ-Bench 4 | — | 1272 |
Frequently asked questions
Is Claude Sonnet 4.5 better than GPT-5.4?
GPT-5.4 is the stronger model overall, scoring 59.4 to 44.1 on the Noometry Index.
Which is cheaper, Claude Sonnet 4.5 or GPT-5.4?
GPT-5.4 is cheaper. It lists at $2.50 per million input tokens and $15 per million output tokens; Claude Sonnet 4.5 lists at $3 and $15.
Is Claude Sonnet 4.5 or GPT-5.4 better for coding?
GPT-5.4 scores higher on coding benchmarks: 52.6 versus 47.3 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 200K.
How many benchmarks do Claude Sonnet 4.5 and GPT-5.4 share?
55 benchmarks have published results for both models. Claude Sonnet 4.5 has 73 scored results on Noometry and GPT-5.4 has 68.