Model comparison
Claude Opus 4.5 vs GPT-5.2
GPT-5.2 is the stronger model overall, scoring 54.1 to 50.5 on the Noometry Index.
Last verified . 65 shared benchmarks.
Summary
- They share 65 benchmarks with published results for both. Claude Opus 4.5 scores higher in 6 categories and GPT-5.2 in 4 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-5.2 leads 60.0 to 38.6.
- The biggest single-benchmark swing is VPCT: 40% for Claude Opus 4.5 and 84% for GPT-5.2.
- GPT-5.2 is cheaper at $1.75 / $14 per million input/output tokens, against $5 / $25 for Claude Opus 4.5.
- GPT-5.2 accepts more context: 400K tokens versus 200K.
Side by side
| Claude Opus 4.5 | GPT-5.2 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 50.5 | 54.1 |
| Released | 2025-11-01 | 2025-12-11 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $5 | $1.75 |
| Output $ / M tokens | $25 | $14 |
| Results tracked | 69 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 4.5 leads
Claude Opus 4.5: 54.8 (#27), GPT-5.2: 51.6 (#37)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| SWE-bench Verified | 76.7% | 73.8% |
| SWE-bench Verified (bash only) | 76.8% | 72.8% |
| LMArena WebDev | 1494 | 1416 |
| SWE-bench Multilingual | 70.7% | 66.7% |
| GSO | 26.5% | 27.4% |
| WeirdML | 63.7% | 72.2% |
| LMArena Coding | 1504 | 1447 |
| ALE-Bench | 1,025 | 1,294 |
| AlgoTune | 1.77 | 2.05 |
Agentic & Tool Use Claude Opus 4.5 leads
Claude Opus 4.5: 47.3 (#12), GPT-5.2: 40.2 (#24)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| Terminal-Bench | 63.1% | 64.9% |
| Berkeley Function Calling Leaderboard | 77.5% | 55.9% |
| GDPval | 45.5% | 49.7% |
| Remote Labor Index | 3.8% | 2.5% |
| τ²-bench Airline | 84% | 83% |
| τ²-bench Banking | 24.7% | 32.2% |
| τ²-bench Retail | 79.6% | 81.6% |
| τ²-bench Telecom | 92.3% | 89.7% |
| DeepResearch Bench | 54.8% | 41.1% |
| LMArena Search | 1180 | 1207 |
| METR Time Horizons | 75% | 75.3% |
| Vending-Bench 2 | 4,967 | 3,591 |
| Cybench | 82% | — |
| OSWorld | 66.3% | — |
| BALROG | 43.5% | — |
Reasoning GPT-5.2 leads
Claude Opus 4.5: 42.6 (#51), GPT-5.2: 50.2 (#35)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| ARC-AGI-2 | 37.6% | 52.9% |
| SimpleBench | 62% | 45.8% |
| Kagi LLM Benchmark | 80.2% | 73.3% |
| NYT Connections (extended) | 52.5% | 83.6% |
| ARC-AGI-1 | 80% | 86.2% |
| Chess Puzzles | 12% | 49% |
| EnigmaEval | 11.9% | 10.4% |
| EBR-Bench | 14.3% | 23% |
| LMArena Hard Prompts | 1476 | 1445 |
| Mystery Game Puzzles | 22% | 23% |
| DTBench | 89.9% | 90.9% |
| LMCA | 44.5% | 43.9% |
| Epoch Capabilities Index | 150.09 | 153.45 |
| ForecastBench | 60.7 | 60.1 |
Math GPT-5.2 leads
Claude Opus 4.5: 38.6 (#132), GPT-5.2: 60.0 (#38)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 34.4% | 67.4% |
| FrontierMath Tier 4 | 4.9% | 31.7% |
| OTIS Mock AIME 2024-2025 | 86.1% | 96.1% |
| ProofBench | 36% | 15% |
| LMArena Math | 1463 | 1440 |
| FrontierMath (Feb 2025 set) | 20.7% | 40.7% |
| FrontierMath Tier 4 (v1) | 4.2% | 18.8% |
| MathArena Final-Answer Competitions | — | 72% |
Knowledge GPT-5.2 leads
Claude Opus 4.5: 56.5 (#44), GPT-5.2: 59.3 (#32)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| GPQA Diamond | 86% | 91.4% |
| Humanity's Last Exam | 25.2% | 27.8% |
| SimpleQA Verified | 45.7% | 37.1% |
| Vectara Hallucination Rate | 10.9% | 8.4% |
| LMArena Expert | 1487 | 1445 |
Multimodal GPT-5.2 leads
Claude Opus 4.5: 31.4 (#107), GPT-5.2: 51.3 (#7)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| VPCT | 40% | 84% |
| Furniture Assembly | 28.3% | 38.3% |
| LMArena Document | 1462 | 1405 |
| LMArena Vision | — | 1268 |
| GeoBench | 75% | — |
Multilingual Too close to call
Claude Opus 4.5: 54.3 (#47), GPT-5.2: 53.4 (#67)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Non-English | 1438 | 1425 |
| LMArena Chinese | 1470 | 1460 |
| LMArena French | 1471 | 1455 |
| LMArena German | 1449 | 1448 |
| LMArena Japanese | 1416 | 1420 |
| LMArena Korean | 1424 | 1392 |
| LMArena Russian | 1447 | 1440 |
| LMArena Spanish | 1458 | 1433 |
Instruction Following Claude Opus 4.5 leads
Claude Opus 4.5: 77.5 (#19), GPT-5.2: 74.7 (#89)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Instruction Following | 1478 | 1417 |
Long Context Claude Opus 4.5 leads
Claude Opus 4.5: 46.5 (#22), GPT-5.2: 44.0 (#78)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| CL-bench | 21.1% | 18.2% |
| LMArena Longer Query | 1480 | 1428 |
Writing & Preference Claude Opus 4.5 leads
Claude Opus 4.5: 68.1 (#28), GPT-5.2: 66.8 (#32)
| Benchmark | Claude Opus 4.5 | GPT-5.2 |
|---|---|---|
| LMArena Text | 1451 | 1439 |
| LMArena Creative Writing | 1445 | 1401 |
| EQ-Bench Creative Writing | 1687 | 1703 |
| LMArena Multi-Turn | 1466 | 1458 |
Frequently asked questions
Is Claude Opus 4.5 better than GPT-5.2?
GPT-5.2 is the stronger model overall, scoring 54.1 to 50.5 on the Noometry Index.
Which is cheaper, Claude Opus 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 Opus 4.5 lists at $5 and $25.
Is Claude Opus 4.5 or GPT-5.2 better for coding?
Claude Opus 4.5 scores higher on coding benchmarks: 54.8 versus 51.6 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 4.5 and GPT-5.2 share?
65 benchmarks have published results for both models. Claude Opus 4.5 has 69 scored results on Noometry and GPT-5.2 has 67.