Model comparison
Claude Sonnet 4.5 vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 44.1 on the Noometry Index.
Last verified . 63 shared benchmarks.
Summary
- They share 63 benchmarks with published results for both. Claude Sonnet 4.5 scores higher in 2 categories and GPT-5.2 in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.2 leads 60.0 to 32.3.
- The biggest single-benchmark swing is NYT Connections (extended): 37.3% for Claude Sonnet 4.5 and 83.6% for GPT-5.2.
- GPT-5.2 is cheaper at $1.75 / $14 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.5.
- GPT-5.2 accepts more context: 400K tokens versus 200K.
Side by side
| Claude Sonnet 4.5 | GPT-5.2 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 44.1 | 54.1 |
| Released | 2025-09-29 | 2025-12-11 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $3 | $1.75 |
| Output $ / M tokens | $15 | $14 |
| Results tracked | 73 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
Claude Sonnet 4.5: 47.3 (#61), GPT-5.2: 51.6 (#37)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| SWE-bench Verified | 71.3% | 73.8% |
| SWE-bench Verified (bash only) | 71.4% | 72.8% |
| LMArena WebDev | 1393 | 1416 |
| SWE-bench Multilingual | 67% | 66.7% |
| GSO | 14.7% | 27.4% |
| WeirdML | 47.7% | 72.2% |
| LMArena Coding | 1489 | 1447 |
| ALE-Bench | 796.15 | 1,294 |
| AlgoTune | 1.52 | 2.05 |
| SciCode | 44.7% | — |
Agentic & Tool Use GPT-5.2 leads
Claude Sonnet 4.5: 38.3 (#32), GPT-5.2: 40.2 (#24)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| Terminal-Bench | 46.5% | 64.9% |
| Berkeley Function Calling Leaderboard | 73.2% | 55.9% |
| GDPval | 42.5% | 49.7% |
| Remote Labor Index | 2.1% | 2.5% |
| τ²-bench Airline | 72% | 83% |
| τ²-bench Banking | 25.3% | 32.2% |
| τ²-bench Retail | 72.4% | 81.6% |
| τ²-bench Telecom | 84.9% | 89.7% |
| DeepResearch Bench | 52.6% | 41.1% |
| LMArena Search | 1159 | 1207 |
| METR Time Horizons | 67.4% | 75.3% |
| Vending-Bench 2 | 3,839 | 3,591 |
| Cybench | 60% | — |
| OSWorld | 62.9% | — |
Reasoning GPT-5.2 leads
Claude Sonnet 4.5: 26.9 (#125), GPT-5.2: 50.2 (#35)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 13.6% | 52.9% |
| SimpleBench | 54.3% | 45.8% |
| Kagi LLM Benchmark | 57.9% | 73.3% |
| NYT Connections (extended) | 37.3% | 83.6% |
| ARC-AGI-1 | 63.7% | 86.2% |
| Chess Puzzles | 12% | 49% |
| EnigmaEval | 6% | 10.4% |
| EBR-Bench | 2.4% | 23% |
| LMArena Hard Prompts | 1462 | 1445 |
| Mystery Game Puzzles | 17% | 23% |
| DTBench | 83.2% | 90.9% |
| LMCA | 38.8% | 43.9% |
| Epoch Capabilities Index | 146.84 | 153.45 |
| ForecastBench | 61.9 | 60.1 |
| CritPt | 1.1% | — |
Math GPT-5.2 leads
Claude Sonnet 4.5: 32.3 (#216), GPT-5.2: 60.0 (#38)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 23.9% | 67.4% |
| FrontierMath Tier 4 | 2.4% | 31.7% |
| OTIS Mock AIME 2024-2025 | 77.8% | 96.1% |
| ProofBench | 19% | 15% |
| LMArena Math | 1449 | 1440 |
| FrontierMath (Feb 2025 set) | 15.2% | 40.7% |
| FrontierMath Tier 4 (v1) | 4.2% | 18.8% |
| MathArena Final-Answer Competitions | — | 72% |
| Omni-MATH | 55.3% | — |
| MATH Level 5 | 97.7% | — |
Knowledge GPT-5.2 leads
Claude Sonnet 4.5: 48.4 (#76), GPT-5.2: 59.3 (#32)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 82.3% | 91.4% |
| Humanity's Last Exam | 13.7% | 27.8% |
| SimpleQA Verified | 30.7% | 37.1% |
| Vectara Hallucination Rate | 12% | 8.4% |
| LMArena Expert | 1482 | 1445 |
| MMLU-Pro | 86.9% | — |
| GPQA (HELM) | 68.6% | — |
Multimodal GPT-5.2 leads
Claude Sonnet 4.5: 34.8 (#89), GPT-5.2: 51.3 (#7)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| VPCT | 39.8% | 84% |
| LMArena Document | 1450 | 1405 |
| LMArena Vision | — | 1268 |
| Furniture Assembly | — | 38.3% |
Multilingual Too close to call
Claude Sonnet 4.5: 53.4 (#69), GPT-5.2: 53.4 (#67)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1425 | 1425 |
| LMArena Chinese | 1459 | 1460 |
| LMArena French | 1458 | 1455 |
| LMArena German | 1427 | 1448 |
| LMArena Japanese | 1390 | 1420 |
| LMArena Korean | 1403 | 1392 |
| LMArena Russian | 1437 | 1440 |
| LMArena Spanish | 1457 | 1433 |
Instruction Following Too close to call
Claude Sonnet 4.5: 75.0 (#78), GPT-5.2: 74.7 (#89)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1459 | 1417 |
| IFEval | 85% | — |
Long Context Claude Sonnet 4.5 leads
Claude Sonnet 4.5: 45.2 (#46), GPT-5.2: 44.0 (#78)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1476 | 1428 |
| CL-bench | — | 18.2% |
Writing & Preference Too close to call
Claude Sonnet 4.5: 66.5 (#34), GPT-5.2: 66.8 (#32)
| Benchmark | Claude Sonnet 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Text | 1439 | 1439 |
| LMArena Creative Writing | 1442 | 1401 |
| EQ-Bench Creative Writing | 1678 | 1703 |
| LMArena Multi-Turn | 1465 | 1458 |
| WildBench | 85.4% | — |
Frequently asked questions
Is Claude Sonnet 4.5 better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 44.1 on the Noometry Index.
Which is cheaper, Claude Sonnet 4.5 or GPT-5.2?
GPT-5.2 is cheaper. It lists at $1.75 per million input tokens and $14 per million output tokens; Claude Sonnet 4.5 lists at $3 and $15.
Is Claude Sonnet 4.5 or GPT-5.2 better for coding?
GPT-5.2 scores higher on coding benchmarks: 51.6 versus 47.3 in the Noometry coding category.
Which has the bigger context window?
GPT-5.2 does, with 400K tokens against 200K.
How many benchmarks do Claude Sonnet 4.5 and GPT-5.2 share?
63 benchmarks have published results for both models. Claude Sonnet 4.5 has 73 scored results on Noometry and GPT-5.2 has 67.