Model comparison
Claude Opus 4 vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 43.1 on the Noometry Index.
Last verified . 43 shared benchmarks.
Summary
- They share 43 benchmarks with published results for both. Claude Opus 4 scores higher in 1 category and GPT-5.2 in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.2 leads 50.2 to 27.3.
- The biggest single-benchmark swing is ARC-AGI-1: 35.7% for Claude Opus 4 and 86.2% for GPT-5.2.
- GPT-5.2 is cheaper at $1.75 / $14 per million input/output tokens, against $15 / $75 for Claude Opus 4.
- GPT-5.2 accepts more context: 400K tokens versus 200K.
Side by side
| Claude Opus 4 | GPT-5.2 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 43.1 | 54.1 |
| Released | 2025-05-22 | 2025-12-11 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 32K | 128K |
| Input $ / M tokens | $15 | $1.75 |
| Output $ / M tokens | $75 | $14 |
| Results tracked | 56 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.2 leads
Claude Opus 4: 47.2 (#62), GPT-5.2: 51.6 (#37)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| SWE-bench Verified | 70.7% | 73.8% |
| SWE-bench Verified (bash only) | 67.6% | 72.8% |
| GSO | 6.9% | 27.4% |
| WeirdML | 43.7% | 72.2% |
| LMArena Coding | 1442 | 1447 |
| AlgoTune | 1.33 | 2.05 |
| Aider Polyglot | 72% | — |
| LMArena WebDev | — | 1416 |
| SWE-bench Multilingual | — | 66.7% |
| ALE-Bench | — | 1,294 |
Agentic & Tool Use GPT-5.2 leads
Claude Opus 4: 34.8 (#42), GPT-5.2: 40.2 (#24)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| DeepResearch Bench | 46.8% | 41.1% |
| LMArena Search | 1127 | 1207 |
| METR Time Horizons | 63.9% | 75.3% |
| Terminal-Bench | — | 64.9% |
| Berkeley Function Calling Leaderboard | — | 55.9% |
| GDPval | — | 49.7% |
| Remote Labor Index | — | 2.5% |
| τ²-bench Airline | — | 83% |
| τ²-bench Banking | — | 32.2% |
| τ²-bench Retail | — | 81.6% |
| τ²-bench Telecom | — | 89.7% |
| Cybench | 38% | — |
| Vending-Bench 2 | — | 3,591 |
Reasoning GPT-5.2 leads
Claude Opus 4: 27.3 (#121), GPT-5.2: 50.2 (#35)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 8.6% | 52.9% |
| SimpleBench | 58.8% | 45.8% |
| Kagi LLM Benchmark | 74.3% | 73.3% |
| ARC-AGI-1 | 35.7% | 86.2% |
| EnigmaEval | 5.6% | 10.4% |
| LMArena Hard Prompts | 1399 | 1445 |
| DTBench | 81.6% | 90.9% |
| LMCA | 37.4% | 43.9% |
| Epoch Capabilities Index | 142.67 | 153.45 |
| ForecastBench | 61.1 | 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 Opus 4: 42.0 (#86), GPT-5.2: 60.0 (#38)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 64.4% | 96.1% |
| LMArena Math | 1390 | 1440 |
| FrontierMath (Feb 2025 set) | 4.5% | 40.7% |
| FrontierMath Tier 4 (v1) | 4.2% | 18.8% |
| FrontierMath (Tiers 1-3) | — | 67.4% |
| FrontierMath Tier 4 | — | 31.7% |
| MathArena Final-Answer Competitions | — | 72% |
| ProofBench | — | 15% |
| Omni-MATH | 61.6% | — |
| MATH Level 5 | 85% | — |
Knowledge GPT-5.2 leads
Claude Opus 4: 44.0 (#88), GPT-5.2: 59.3 (#32)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 76.3% | 91.4% |
| Humanity's Last Exam | 10.7% | 27.8% |
| Vectara Hallucination Rate | 12% | 8.4% |
| LMArena Expert | 1386 | 1445 |
| SimpleQA Verified | — | 37.1% |
| MMLU-Pro | 87.5% | — |
| Confabulations | 15.9% | — |
| GPQA (HELM) | 70.8% | — |
Multimodal GPT-5.2 leads
Claude Opus 4: 31.5 (#106), GPT-5.2: 51.3 (#7)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| LMArena Vision | 1192 | 1268 |
| VPCT | 38% | 84% |
| GeoBench | 49% | — |
| Furniture Assembly | — | 38.3% |
| LMArena Document | — | 1405 |
Multilingual GPT-5.2 leads
Claude Opus 4: 48.8 (#138), GPT-5.2: 53.4 (#67)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1362 | 1425 |
| LMArena Chinese | 1386 | 1460 |
| LMArena French | 1372 | 1455 |
| LMArena German | 1391 | 1448 |
| LMArena Japanese | 1331 | 1420 |
| LMArena Korean | 1321 | 1392 |
| LMArena Russian | 1392 | 1440 |
| LMArena Spanish | 1389 | 1433 |
Instruction Following Claude Opus 4 leads
Claude Opus 4: 77.1 (#28), GPT-5.2: 74.7 (#89)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1406 | 1417 |
| IFEval | 91.8% | — |
Long Context GPT-5.2 leads
Claude Opus 4: 39.6 (#172), GPT-5.2: 44.0 (#78)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| LMArena Longer Query | 1422 | 1428 |
| Fiction.LiveBench | 61.1% | — |
| CL-bench | — | 18.2% |
Writing & Preference GPT-5.2 leads
Claude Opus 4: 61.2 (#89), GPT-5.2: 66.8 (#32)
| Benchmark | Claude Opus 4 | GPT-5.2 |
|---|---|---|
| LMArena Text | 1377 | 1439 |
| LMArena Creative Writing | 1387 | 1401 |
| EQ-Bench Creative Writing | 1580 | 1703 |
| LMArena Multi-Turn | 1396 | 1458 |
| Short-Story Creative Writing | 83.6% | — |
| WildBench | 85.2% | — |
Frequently asked questions
Is Claude Opus 4 better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 43.1 on the Noometry Index.
Which is cheaper, Claude Opus 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 Opus 4 lists at $15 and $75.
Is Claude Opus 4 or GPT-5.2 better for coding?
GPT-5.2 scores higher on coding benchmarks: 51.6 versus 47.2 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 Opus 4 and GPT-5.2 share?
43 benchmarks have published results for both models. Claude Opus 4 has 56 scored results on Noometry and GPT-5.2 has 67.