General Agents Benchmark

SATA-Bench

SATA-Bench: structured-output attribute/typed answer evaluation (case exact accuracy), scored on a 0-100 scale in Cloudflare's internal Decision Index 0.2.1 suite.

Category

General Agents

Score type

percent

Scale

100%

higher is better

Models tested

2

What is SATA-Bench?

SATA-Bench is a benchmark for how well a model acts as a general-purpose agent. The official description says: "SATA-Bench: structured-output attribute/typed answer evaluation (case exact accuracy), scored on a 0-100 scale in Cloudflare's internal Decision Index 0.2.1 suite". Below we describe in simple words what benchmarks of this type generally examine.

What does it test?

SATA-Bench measures how well a model acts as a general-purpose agent. In detail, it checks:

  • solving long, multi-step tasks with many intermediate decisions
  • using tools correctly (browsers, terminals, APIs, calendars, other apps)
  • adapting when something unexpected happens mid-task
  • staying on goal over long task horizons instead of wandering off

How to read the score

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.

Official methodology note: SATA-Bench: structured-output attribute/typed answer evaluation (case exact accuracy), scored on a 0-100 scale in Cloudflare's internal Decision Index 0.2.1 suite.

Frequently asked questions

What is the SATA-Bench benchmark?

SATA-Bench is a benchmark for how well a model acts as a general-purpose agent. The official description says: 'SATA-Bench: structured-output attribute/typed answer evaluation (case exact accuracy), scored on a 0-100 scale in Cloudflare's internal Decision Index 0.2.1 suite'. Below we describe in simple words what benchmarks of this type generally examine.

What does SATA-Bench test?

SATA-Bench mainly measures how well a model acts as a general-purpose agent. Concretely, it checks: solving long, multi-step tasks with many intermediate decisions; using tools correctly (browsers, terminals, APIs, calendars, other apps); adapting when something unexpected happens mid-task.

How is the SATA-Bench score calculated?

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better. The declared maximum score of this benchmark is 100.0.

Where do I get SATA-Bench data and scores?

The official source is https://clef-evals.workers-ai-mle.workers.dev/. The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/sata-bench/scores) or pull it directly from the Open LLM Radar API with any HTTP client.

Model scores 2 models tested

Reset
# Model Score Date Source Verified
1 Clef-Flash
Cloudflare
36.7%
— Report Verified
2 Clef
Cloudflare
33.8%
— Report Verified

What now?

You've seen how models score on SATA-Bench — here's what to do next.