DRACO

DRACO is a benchmark for how well a model acts as a general-purpose agent. It was built to compare language models on a clearly defined set of tasks under the same grading rules, so scores are directly comparable between models.

DRACO 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

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

What is the DRACO benchmark?

DRACO is a benchmark for how well a model acts as a general-purpose agent. It was built to compare language models on a clearly defined set of tasks under the same grading rules, so scores are directly comparable between models.

What does DRACO test?

DRACO 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 DRACO 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 DRACO data and scores?

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/draco/scores) or pull it directly from the Open LLM Radar API with any HTTP client.