MATH-500
Reported model scores on MATH-500, best parseable score first. Each row keeps its verbatim score, test conditions and provenance, and links to the source coverage it was extracted from. Benchmark profile →
| Model | Score | Conditions | Provenance | Measured | Source |
|---|---|---|---|---|---|
| Qwen3.5-0.8B (fixed by GPT-6 Astra weight-space edit) | 48.1% | GPT-6 Astra fixes Qwen3.5-0.8B by editing 0.016% of its weights (no gradient-based training or environment feedback) | independent | — | coverage → |
| Qwen3.5-0.8B (fixed by GPT-6 Astra weight-space edit) | 7.9% | after editing 0.016% of weights by GPT-6 Astra, without gradient-based training or environment feedback (reported as the starting accuracy before improvement) | independent | — | coverage → |
Scores are only comparable within one benchmark under matching conditions — results under different test setups, and scores from other benchmarks, are not directly comparable.