Model comparison
Claude Opus 4.6 vs Claude Opus 5
Claude Opus 5 is the stronger model overall, scoring 67.8 to 58.2 on the Noometry Index.
Last verified . 46 shared benchmarks.
Summary
- They share 46 benchmarks with published results for both. Claude Opus 4.6 scores higher in 2 categories and Claude Opus 5 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Opus 5 leads 86.2 to 63.0.
- The biggest single-benchmark swing is ProofBench: 50% for Claude Opus 4.6 and 99% for Claude Opus 5.
- Both cost about the same: $5 input and $25 output per million tokens.
Side by side
| Claude Opus 4.6 | Claude Opus 5 | |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Noometry Index | 58.2 | 67.8 |
| Released | 2026-02-04 | 2026-07-24 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1M |
| Max output | 128K | 128K |
| Input $ / M tokens | $5 | $5 |
| Output $ / M tokens | $25 | $25 |
| Results tracked | 68 | 57 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Opus 5 leads
Claude Opus 4.6: 57.2 (#20), Claude Opus 5: 67.5 (#5)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| FrontierCode | 26.6% | 53.4% |
| LMArena WebDev | 1547 | 1691 |
| WeirdML | 78% | 91.8% |
| LMArena Coding | 1536 | 1534 |
| ALE-Bench | 996.5 | 2,165 |
| SWE-bench Verified | 78.7% | — |
| DeepSWE | — | 73.6% |
| SWE-bench Verified (bash only) | 75.6% | — |
| CursorBench | — | 46.6% |
| SWE-bench Multilingual | 72% | — |
| FrontierSWE | — | 52% |
| SciCode | — | 56.4% |
| GSO | 41.2% | — |
| AlgoTune | 1.47 | — |
Agentic & Tool Use Claude Opus 5 leads
Claude Opus 4.6: 51.1 (#4), Claude Opus 5: 55.6 (#1)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| APEX-Agents | 46.3% | 65.8% |
| τ²-bench Banking | 27.3% | 48.7% |
| GBAEval | 44.1% | 79.6% |
| Vending-Bench 2 | 8,018 | 11,182 |
| Terminal-Bench | 79.8% | — |
| OSWorld 2.0 | — | 31.4% |
| Remote Labor Index | 4.2% | — |
| Cybench | 93% | — |
| DeepResearch Bench | 55.3% | — |
| PostTrainBench | — | 35% |
| BALROG | — | 63.4% |
| GDP.pdf | — | 24% |
| LMArena Search | 1253 | — |
| METR Time Horizons | 78.9% | — |
Reasoning Claude Opus 5 leads
Claude Opus 4.6: 57.8 (#23), Claude Opus 5: 77.2 (#4)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| ARC-AGI-2 | 69.2% | 90.4% |
| SimpleBench | 67.6% | 80.6% |
| NYT Connections (extended) | 92.1% | 94.3% |
| ARC-AGI-1 | 94% | 97.5% |
| Chess Puzzles | 17% | 42% |
| EBR-Bench | 12.7% | 45.7% |
| LMArena Hard Prompts | 1527 | 1526 |
| Mystery Game Puzzles | 25% | 59% |
| DTBench | 91.2% | 97.9% |
| LMCA | 55.8% | 64.5% |
| Epoch Capabilities Index | 155.24 | 162.78 |
| Kagi LLM Benchmark | 83.6% | — |
| CritPt | — | 29.1% |
| EnigmaEval | 7.6% | — |
| Thematic Generalization | 80.6% | — |
| Bench to the Future 3 | — | 0.12 |
| ForecastBench | 60 | — |
Math Claude Opus 5 leads
Claude Opus 4.6: 63.0 (#31), Claude Opus 5: 86.2 (#8)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 66% | 85.6% |
| FrontierMath Tier 4 | 26.8% | 73.2% |
| OTIS Mock AIME 2024-2025 | 94.4% | 98.9% |
| ProofBench | 50% | 99% |
| LMArena Math | 1519 | 1531 |
| MathArena Final-Answer Competitions | 78.5% | — |
| FrontierMath (Feb 2025 set) | 40.7% | — |
| FrontierMath Tier 4 (v1) | 22.9% | — |
Knowledge Claude Opus 5 leads
Claude Opus 4.6: 61.9 (#26), Claude Opus 5: 66.8 (#9)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| GPQA Diamond | 90.5% | 93.9% |
| SimpleQA Verified | 47% | 59.9% |
| LMArena Expert | 1546 | 1557 |
| Humanity's Last Exam | 34.4% | — |
| Vectara Hallucination Rate | 12.2% | — |
Multimodal Claude Opus 5 leads
Claude Opus 4.6: 37.3 (#74), Claude Opus 5: 50.8 (#8)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| LMArena Vision | 1316 | 1319 |
| Furniture Assembly | 28.3% | 60.8% |
| LMArena Document | 1507 | 1516 |
| Blueprint-Bench 2 | — | 30.4% |
Multilingual Too close to call
Claude Opus 4.6: 57.9 (#6), Claude Opus 5: 58.8 (#4)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| LMArena Non-English | 1489 | 1501 |
| LMArena Chinese | 1551 | 1574 |
| LMArena French | 1513 | 1519 |
| LMArena German | 1502 | 1524 |
| LMArena Japanese | 1484 | 1516 |
| LMArena Korean | 1464 | 1521 |
| LMArena Russian | 1497 | 1507 |
| LMArena Spanish | 1510 | 1519 |
Instruction Following Too close to call
Claude Opus 4.6: 79.5 (#4), Claude Opus 5: 79.2 (#7)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| LMArena Instruction Following | 1523 | 1517 |
Long Context Claude Opus 4.6 leads
Claude Opus 4.6: 48.1 (#13), Claude Opus 5: 46.5 (#21)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| LMArena Longer Query | 1520 | 1515 |
| CL-bench | 20.7% | — |
| CL-bench Life | 17% | — |
Writing & Preference Claude Opus 5 leads
Claude Opus 4.6: 73.5 (#10), Claude Opus 5: 79.2 (#1)
| Benchmark | Claude Opus 4.6 | Claude Opus 5 |
|---|---|---|
| LMArena Text | 1503 | 1507 |
| LMArena Creative Writing | 1505 | 1491 |
| EQ-Bench Creative Writing | 1809 | 2133 |
| EQ-Bench 4 | 1223 | 1385 |
| LMArena Multi-Turn | 1513 | 1499 |
Frequently asked questions
Is Claude Opus 4.6 better than Claude Opus 5?
Claude Opus 5 is the stronger model overall, scoring 67.8 to 58.2 on the Noometry Index.
Which is cheaper, Claude Opus 4.6 or Claude Opus 5?
Claude Opus 5 is cheaper. It lists at $5 per million input tokens and $25 per million output tokens; Claude Opus 4.6 lists at $5 and $25.
Is Claude Opus 4.6 or Claude Opus 5 better for coding?
Claude Opus 5 scores higher on coding benchmarks: 67.5 versus 57.2 in the Noometry coding category.
Which has the bigger context window?
Both accept 1M tokens.
How many benchmarks do Claude Opus 4.6 and Claude Opus 5 share?
46 benchmarks have published results for both models. Claude Opus 4.6 has 68 scored results on Noometry and Claude Opus 5 has 57.