SlopCodeBench (Plan First): leaderboard
Metric: Strict solve rate (%): share of the 196 checkpoints whose workspace passes every test, including all earlier checkpoints' regression tests (missing checkpoints count as unsolved), on SlopCodeBench's 36 iterative problems (196 checkpoints): the agent extends its own prior code as the CLI or API specification evolves, with hidden tests, run in the model's native coding harness (the harness and reasoning level are in the label), using a prompt requiring a written plan before coding; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 (High) | 9.7 | #76 (GPT-5.4) |
| 2 | GPT-5.5 (High) | 8.2 | #26 (GPT-5.5) |
| 3 | GPT-5.3 Codex (High) | 8.2 | #68 (GPT-5.3 Codex) |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=slopcodebench-plan-first · How It Works · Data refreshed daily, snapshot 2026-10-11.