Across 11 shared benchmarks, Claude Sonnet 4.5 scores higher on 8 and Granite 4.2 8B on 3. The widest gap is τ³-Bench, where Granite 4.2 8B scores 66.3 against 19.
| Benchmark | Claude Sonnet 4.5 | Granite 4.2 8B |
|---|---|---|
| AIME25 no tools | 88 | 86.7 |
| gdpval | 40.4 | 1189 |
| GPQA Diamond | 83.4 | 64.1 |
| LiveCodeBench v6 | 64 | 73.2 |
| MMLU-Pro | 88.2 | 74 |
| scicode | 45 | 36.1 |
| SWE-bench Multilingual | 68 | 30.8 |
| SWE-bench Pro | 43.6 | 19.1 |
| SWE-bench Verified | 82 | 47.7 |
| Terminal-Bench 2.1 | 55.8 | 20.6 |
| τ³-Bench | 19 | 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.