SciVisAgentBench - Object Identification: leaderboard
Metric: Overall score (0-100) on SciVisAgentBench's object identification questions with discrete answers checked by rule-based verifiers: a Claude-Opus-4.6 judge rates the agent's output against the expert ground truth on expert-authored rubrics (rule-based checks where available), normalized by completion; mean of three trials; higher is better. Source: arxiv.org. Saturation forecast: Around May 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 43.33 | #105 |
| 2 | Claude Sonnet 4.5 | 41.5 | #138 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=scivisagentbench-object-identification · How It Works · Data refreshed daily, snapshot 2026-10-11.