Model comparison
Claude Sonnet 4 vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 40.8 on the Noometry Index.
Last verified . 41 shared benchmarks.
Summary
- They share 41 benchmarks with published results for both. Claude Sonnet 4 scores higher in 0 categories and GPT-5.2 in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.2 leads 50.2 to 22.9.
- The biggest single-benchmark swing is VPCT: 34% for Claude Sonnet 4 and 84% 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.
- GPT-5.2 accepts more context: 400K tokens versus 200K.
Side by side
| Claude Sonnet 4 | GPT-5.2 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 40.8 | 54.1 |
| Released | 2025-05-22 | 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 | 58 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
Claude Sonnet 4: 43.5 (#88), GPT-5.2: 51.6 (#37)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| SWE-bench Verified (bash only) | 64.9% | 72.8% |
| GSO | 4.9% | 27.4% |
| WeirdML | 46.1% | 72.2% |
| LMArena Coding | 1414 | 1447 |
| ALE-Bench | 655.35 | 1,294 |
| SWE-bench Verified | — | 73.8% |
| Aider Polyglot | 61.3% | — |
| LMArena WebDev | — | 1416 |
| SWE-bench Multilingual | — | 66.7% |
| SciCode | 40% | — |
| AlgoTune | — | 2.05 |
Agentic & Tool Use GPT-5.2 leads
Claude Sonnet 4: 38.5 (#31), GPT-5.2: 40.2 (#24)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| DeepResearch Bench | 46.6% | 41.1% |
| METR Time Horizons | 62% | 75.3% |
| Terminal-Bench | — | 64.9% |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| GDPval | — | 49.7% |
| Remote Labor Index | — | 2.5% |
| TheAgentCompany | 33.1% | — |
| τ²-bench Airline | — | 83% |
| τ²-bench Banking | — | 32.2% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| Cybench | 35% | — |
| OSWorld | 43.9% | — |
| LMArena Search | — | 1207 |
| Vending-Bench 2 | — | 3,591 |
Reasoning GPT-5.2 leads
Claude Sonnet 4: 22.9 (#187), GPT-5.2: 50.2 (#35)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 5.9% | 52.9% |
| SimpleBench | 45.5% | 45.8% |
| Kagi LLM Benchmark | 73% | 73.3% |
| ARC-AGI-1 | 40% | 86.2% |
| EnigmaEval | 3.1% | 10.4% |
| LMArena Hard Prompts | 1372 | 1445 |
| DTBench | 77.1% | 90.9% |
| LMCA | 29% | 43.9% |
| Epoch Capabilities Index | 141.69 | 153.45 |
| ForecastBench | 60.2 | 60.1 |
| NYT Connections (extended) | — | 83.6% |
| CritPt | 0.3% | — |
| Chess Puzzles | — | 49% |
| EBR-Bench | — | 23% |
| Mystery Game Puzzles | — | 23% |
Math GPT-5.2 leads
Claude Sonnet 4: 43.3 (#80), GPT-5.2: 60.0 (#38)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 96.1% |
| LMArena Math | 1375 | 1440 |
| FrontierMath (Feb 2025 set) | 4.1% | 40.7% |
| FrontierMath Tier 4 (v1) | 0% | 18.8% |
| FrontierMath (Tiers 1-3) | — | 67.4% |
| FrontierMath Tier 4 | — | 31.7% |
| MathArena Final-Answer Competitions | — | 72% |
| ProofBench | — | 15% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
Knowledge GPT-5.2 leads
Claude Sonnet 4: 41.8 (#108), GPT-5.2: 59.3 (#32)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 79.2% | 91.4% |
| Humanity's Last Exam | 7.8% | 27.8% |
| Vectara Hallucination Rate | 10.3% | 8.4% |
| LMArena Expert | 1372 | 1445 |
| SimpleQA Verified | — | 37.1% |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal GPT-5.2 leads
Claude Sonnet 4: 26.2 (#121), GPT-5.2: 51.3 (#7)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| LMArena Vision | 1191 | 1268 |
| VPCT | 34% | 84% |
| GeoBench | 37% | — |
| Furniture Assembly | — | 38.3% |
| LMArena Document | — | 1405 |
| MindCube | 44.8% | — |
Multilingual GPT-5.2 leads
Claude Sonnet 4: 46.7 (#156), GPT-5.2: 53.4 (#67)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1333 | 1425 |
| LMArena Chinese | 1350 | 1460 |
| LMArena French | 1363 | 1455 |
| LMArena German | 1331 | 1448 |
| LMArena Japanese | 1302 | 1420 |
| LMArena Korean | 1291 | 1392 |
| LMArena Russian | 1355 | 1440 |
| LMArena Spanish | 1357 | 1433 |
Instruction Following GPT-5.2 leads
Claude Sonnet 4: 71.7 (#145), GPT-5.2: 74.7 (#89)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1376 | 1417 |
| IFEval | 84% | — |
Long Context GPT-5.2 leads
Claude Sonnet 4: 33.7 (#259), GPT-5.2: 44.0 (#78)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1398 | 1428 |
| Fiction.LiveBench | 46.9% | — |
| CL-bench | — | 18.2% |
Writing & Preference GPT-5.2 leads
Claude Sonnet 4: 57.1 (#132), GPT-5.2: 66.8 (#32)
| Benchmark | Claude Sonnet 4 | GPT-5.2 |
|---|---|---|
| LMArena Text | 1351 | 1439 |
| LMArena Creative Writing | 1345 | 1401 |
| EQ-Bench Creative Writing | 1483 | 1703 |
| LMArena Multi-Turn | 1376 | 1458 |
| Short-Story Creative Writing | 81.4% | — |
| WildBench | 83.8% | — |
Frequently asked questions
Is Claude Sonnet 4 better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 40.8 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 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 lists at $3 and $15.
Is Claude Sonnet 4 or GPT-5.2 better for coding?
GPT-5.2 scores higher on coding benchmarks: 51.6 versus 43.5 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 and GPT-5.2 share?
41 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and GPT-5.2 has 67.