BENCHMARK EVIDENCE / ACTIVE
AutomationBench
professional workflows. Metric: task-success. Direction: higher-is-better. Results below stay separated by evaluator evidence type and comparable group.
1observations
OpenAIevaluator
activestatus
2026-10-04verified
METHOD
Read the score with its configuration.
Scores from different comparable groups are intentionally not merged into one ranking. Reasoning effort, tools, fallback behavior, evaluator and benchmark version stay attached to every observation.
- Provider-reported launch evaluation. Do not merge numerically with AutomationBench-AA without a documented equivalence.
vendor-reported
1 observationopenai-astra-launch-table
| Model | Score | Reasoning | Tools | Source |
|---|---|---|---|---|
| GPT-6 Astra | 41.4% | not published | provider-defined | Evidence ↗ |