claude-opus-4-6 vs gpt-5.2-codex
Both models ran the same corpus under the same verifier — claude and opencode respectively. The harness differs where the names differ, and that is part of what separates them.
claude-opus-4-6 leads by 23.8 points, and the intervals do not overlap.
Where they differ
| Weakness | claude-opus-4-6 | gpt-5.2-codex | Difference |
|---|---|---|---|
| CWE-125Out-of-bounds Read | 69.3% | 41.3% | +28.0 |
| CWE-787Out-of-bounds Write | 50.0% | 41.2% | +8.8 |
| CWE-416Use After Free | 63.6% | 30.8% | +32.8 |
| CWE-908Use of Uninitialized Resource | 11.1% | 22.2% | -11.1 |
| CWE-415Double Free | 66.7% | 50.0% | +16.7 |
| CWE-1284Improper Validation of Specified Quantity in Input | 50.0% | 50.0% | 0.0 |
| CWE-843Access of Resource Using Incompatible Type ('Type Confusion') | 50.0% | 0.0% | +50.0 |
| CWE-476NULL Pointer Dereference | 50.0% | 0.0% | +50.0 |
| CWE-475 | 100.0% | 0.0% | +100.0 |
| CWE-562 | 100.0% | 100.0% | 0.0 |
Per-class samples are small; a difference here is a hint about where to look, not a result. The corpus-level intervals above are the honest comparison.
Corpus cve-bench-136 · run February 2026 · How we test
See the benchmark →