LabOSBench - Post-processing & Completion: leaderboard
Post-processing and completion operations in LabOSBench (Zou et al., 2026), such as saving and exporting images and results, reconstructing data, opening analysis views and returning the instrument to its final state, across its web-based scientific-instrument simulators. Each subtask starts from its canonical pre-state with a 50-step budget and runs twice; the score is the percent of successful runs averaged over the category's subtasks; higher is better.
Metric: Subtask Success Rate (%; mean over the category's operation-level subtasks, 2 runs each). Source: arxiv.org. Saturation forecast: Around July 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Seed-1.6 | 81.2 | #447 |
| 2 | GPT-5.5 | 75 | #48 |
| 3 | Claude Sonnet 4.5 | 68.8 | #244 |
| 4 | Kimi K2.5 | 62.5 | #252 |
| 5 | Claude Opus 4.5 | 62.5 | #152 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=labosbench-post-processing-completion · How It Works · Data refreshed daily, snapshot 2026-11-12.