Across 10 shared benchmarks, Claude Opus 4.6 scores higher on 9 and Granite 4.2 8B on 1. The widest gap is SWE-bench Pro, where Claude Opus 4.6 scores 57.3 against 19.1.
| Benchmark | Claude Opus 4.6 | Granite 4.2 8B |
|---|---|---|
| AIME25 no tools | 95.6 | 86.7 |
| gdpval | 55.9 | 1189 |
| GPQA Diamond | 91.3 | 64.1 |
| LiveCodeBench v6 | 88.8 | 73.2 |
| MMLU-Pro | 89.1 | 74 |
| scicode | 52 | 36.1 |
| SWE-bench Multilingual | 77.8 | 30.8 |
| SWE-bench Pro | 57.3 | 19.1 |
| SWE-bench Verified | 80.8 | 47.7 |
| τ³-Bench | 72.4 | 66.3 |
Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.