Model comparison
Claude Opus 5 vs GPT-5.5
Claude Opus 5 is the stronger model overall, scoring 67.8 to 63.4 on the Noometry Index.
Last verified . 54 shared benchmarks.
Summary
- They share 54 benchmarks with published results for both. Claude Opus 5 scores higher in 9 categories and GPT-5.5 in 1 category; 10 gaps are clear of the uncertainty.
- The widest gap is in coding, where Claude Opus 5 leads 67.5 to 58.2.
- The biggest single-benchmark swing is ProofBench: 99% for Claude Opus 5 and 50% for GPT-5.5.
- Claude Opus 5 is cheaper at $5 / $25 per million input/output tokens, against $5 / $30 for GPT-5.5.
- GPT-5.5 accepts more context: 1.05M tokens versus 1M.
Side by side
| Claude Opus 5 | GPT-5.5 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 67.8 | 63.4 |
| Released | 2026-07-24 | 2026-04-23 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $5 | $5 |
| Output $ / M tokens | $25 | $30 |
| Results tracked | 57 | 71 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 5 leads
Claude Opus 5: 67.5 (#5), GPT-5.5: 58.2 (#17)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| DeepSWE | 73.6% | 67% |
| FrontierCode | 53.4% | 43% |
| LMArena WebDev | 1691 | 1513 |
| SciCode | 56.4% | 56.1% |
| WeirdML | 91.8% | 84.9% |
| LMArena Coding | 1534 | 1494 |
| ALE-Bench | 2,165 | 1,943 |
| SWE-bench Verified | — | 80.6% |
| CursorBench | 46.6% | — |
| FrontierSWE | 52% | — |
| GSO | — | 40.2% |
| MirrorCode | — | 10% |
Agentic & Tool Use Claude Opus 5 leads
Claude Opus 5: 55.6 (#1), GPT-5.5: 50.7 (#6)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| APEX-Agents | 65.8% | 55.1% |
| OSWorld 2.0 | 31.4% | 13% |
| τ²-bench Banking | 48.7% | 44.6% |
| PostTrainBench | 35% | 27.2% |
| GBAEval | 79.6% | 53.2% |
| GDP.pdf | 24% | 26% |
| Vending-Bench 2 | 11,182 | 7,524 |
| Terminal-Bench | — | 84.7% |
| Remote Labor Index | — | 6.3% |
| DeepResearch Bench | — | 54% |
| BALROG | 63.4% | — |
| ExploitBench | — | 47.4% |
| LMArena Search | — | 1242 |
Reasoning Claude Opus 5 leads
Claude Opus 5: 77.2 (#4), GPT-5.5: 72.8 (#11)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| ARC-AGI-2 | 90.4% | 85% |
| SimpleBench | 80.6% | 69% |
| NYT Connections (extended) | 94.3% | 96.2% |
| ARC-AGI-1 | 97.5% | 95% |
| CritPt | 29.1% | 27.1% |
| Chess Puzzles | 42% | 54% |
| EBR-Bench | 45.7% | 34.3% |
| LMArena Hard Prompts | 1526 | 1489 |
| Mystery Game Puzzles | 59% | 56% |
| DTBench | 97.9% | 96% |
| LMCA | 64.5% | 54.3% |
| Bench to the Future 3 | 0.12 | 0.14 |
| Epoch Capabilities Index | 162.78 | 159.1 |
| Kagi LLM Benchmark | — | 88.8% |
| Surface Evolver Bench | — | 88.1% |
| ForecastBench | — | 60.6 |
Math Claude Opus 5 leads
Claude Opus 5: 86.2 (#8), GPT-5.5: 81.7 (#11)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 85.6% | 85.3% |
| FrontierMath Tier 4 | 73.2% | 72.5% |
| OTIS Mock AIME 2024-2025 | 98.9% | 100% |
| ProofBench | 99% | 50% |
| LMArena Math | 1531 | 1486 |
| MathArena Final-Answer Competitions | — | 94.3% |
| FrontierMath (Feb 2025 set) | — | 51.7% |
| FrontierMath Erdős | — | 0% |
| FrontierMath Tier 4 (v1) | — | 35.4% |
Knowledge Claude Opus 5 leads
Claude Opus 5: 66.8 (#9), GPT-5.5: 64.4 (#17)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| GPQA Diamond | 93.9% | 94% |
| SimpleQA Verified | 59.9% | 63% |
| LMArena Expert | 1557 | 1508 |
| Vectara Hallucination Rate | — | 9.3% |
Multimodal Claude Opus 5 leads
Claude Opus 5: 50.8 (#8), GPT-5.5: 46.9 (#12)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| LMArena Vision | 1319 | 1297 |
| Blueprint-Bench 2 | 30.4% | 36.2% |
| Furniture Assembly | 60.8% | 44.2% |
| LMArena Document | 1516 | 1486 |
Multilingual Claude Opus 5 leads
Claude Opus 5: 58.8 (#4), GPT-5.5: 56.4 (#20)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| LMArena Non-English | 1501 | 1467 |
| LMArena Chinese | 1574 | 1533 |
| LMArena French | 1519 | 1486 |
| LMArena German | 1524 | 1480 |
| LMArena Japanese | 1516 | 1498 |
| LMArena Korean | 1521 | 1460 |
| LMArena Russian | 1507 | 1473 |
| LMArena Spanish | 1519 | 1468 |
Instruction Following Claude Opus 5 leads
Claude Opus 5: 79.2 (#7), GPT-5.5: 77.5 (#18)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| LMArena Instruction Following | 1517 | 1479 |
Long Context GPT-5.5 leads
Claude Opus 5: 46.5 (#21), GPT-5.5: 48.3 (#12)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| LMArena Longer Query | 1515 | 1484 |
| CL-bench Life | — | 22.2% |
Writing & Preference Claude Opus 5 leads
Claude Opus 5: 79.2 (#1), GPT-5.5: 72.7 (#13)
| Benchmark | Claude Opus 5 | GPT-5.5 |
|---|---|---|
| LMArena Text | 1507 | 1472 |
| LMArena Creative Writing | 1491 | 1455 |
| EQ-Bench Creative Writing | 2133 | 1844 |
| EQ-Bench 4 | 1385 | 1315 |
| LMArena Multi-Turn | 1499 | 1476 |
Frequently asked questions
Is Claude Opus 5 better than GPT-5.5?
Claude Opus 5 is the stronger model overall, scoring 67.8 to 63.4 on the Noometry Index.
Which is cheaper, Claude Opus 5 or GPT-5.5?
Claude Opus 5 is cheaper. It lists at $5 per million input tokens and $25 per million output tokens; GPT-5.5 lists at $5 and $30.
Is Claude Opus 5 or GPT-5.5 better for coding?
Claude Opus 5 scores higher on coding benchmarks: 67.5 versus 58.2 in the Noometry coding category.
Which has the bigger context window?
GPT-5.5 does, with 1.05M tokens against 1M.
How many benchmarks do Claude Opus 5 and GPT-5.5 share?
54 benchmarks have published results for both models. Claude Opus 5 has 57 scored results on Noometry and GPT-5.5 has 71.