# Results summary (auto-generated by harness/run_all.py)

| Task | chatgpt passes | claude passes |
|---|---|---|
| T01-semver-ranges | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T02-bugfix-venue | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T03-js-csv-parser | 1/1 (score 1) | 1/1 (score 1) |
| T04-sql-streaks | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T05-refactor-shipping | 0/1 (score 0.833) | 1/1 (score 1.0) |
| T06-interval-scheduling | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T07-data-mrr | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T08-logic-puzzle | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T09-js-maplimit | 1/1 (score 1) | 1/1 (score 1) |
| T10-business-hours | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T11-release-notes | 1/1 (score 1.0) | 1/1 (score 1.0) |
| T12-instruction-following | 1/1 (score 1.0) | 0/1 (score 0.909) |
| T13-predict-output | 0/1 (score 0.818) | 1/1 (score 1.0) |

- **chatgpt**: pass@1 0.846 over 13 attempts; tasks passed on every run: 0/13; mean partial score 0.973; pass rate by run {1: 0.846}; median seconds 17
- **claude**: pass@1 0.923 over 13 attempts; tasks passed on every run: 0/13; mean partial score 0.993; pass rate by run {1: 0.923}; median seconds 27.5
- Difference chatgpt - claude in pass@1: -0.077 (bootstrap 95% CI over tasks [-0.308, 0.154]). If the interval includes 0, call it a tie.