reasoning benchmark
AutomationBench-AA
AutomationBench-AA is Artificial Analysis's independently run version of AutomationBench, covering 657 real-world SaaS workflow tasks across 40 simulated applications (e.g. Gmail, Slack, Salesforce, HubSpot). It scores the share of objectives an agent completes without violating business guardrails, using a private held-out task set.
Updated Aug 11, 2026
Models1
Model coverage1
MetricScore
EvidenceB
AutomationBench-AA Ranking
Higher score ranks better on this benchmark.
| 1 | XA | 51.4% | 100.0% | 1 | B |
AutomationBench-AA Highlights
The leading models and scores on this benchmark.
What is AutomationBench-AA?
What AutomationBench-AA measures and how its scores work.
AutomationBench-AA is Artificial Analysis's independently run version of AutomationBench, covering 657 real-world SaaS workflow tasks across 40 simulated applications (e.g. Gmail, Slack, Salesforce, HubSpot). It scores the share of objectives an agent completes without violating business guardrails, using a private held-out task set.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- AutomationBench-AA
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- automationbench-aa|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about AutomationBench-AA.
Which model scores highest on AutomationBench-AA?
Grok 4.5 is currently ranked first with 51.4%.
What does AutomationBench-AA measure?
AutomationBench-AA is Artificial Analysis's independently run version of AutomationBench, covering 657 real-world SaaS workflow tasks across 40 simulated applications (e.g. Gmail, Slack, Salesforce, HubSpot). It scores the share of objectives an agent completes without violating business guardrails, using a private held-out task set.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.