Task samples

Environments

Get in touch

← Environments

Second Opinion

Second Opinion

Second Opinion

A clinical care environment where agents review patient records and imaging, document decisions, place orders, and coordinate follow-up.

A clinical care environment where agents review patient records and imaging, document decisions, place orders, and coordinate follow-up.

The workspace

The workspace

The workspace

Agents work with clinical records, documents, and imaging, then write notes, place orders, and arrange follow-up. The tools available depend on the task; changes are saved in the working patient chart.

Agents work with clinical records, documents, and imaging, then write notes, place orders, and arrange follow-up. The tools available depend on the task; changes are saved in the working patient chart.

The assignment

The assignment

The assignment

Second Opinion includes six clinical task definitions. The ankle injury study includes 52 scored attempts; results for the other tasks are not presented.

Second Opinion includes six clinical task definitions. The ankle injury study includes 52 scored attempts; results for the other tasks are not presented.

The evidence

The evidence

The evidence

The ankle injury study checks the saved assessment, care arrangements, earlier history, imaging decisions, and preservation of existing clinical records.

The ankle injury study checks the saved assessment, care arrangements, earlier history, imaging decisions, and preservation of existing clinical records.

Second Opinion · Clinical care environment

Assess an ankle injury. Complete the care.

Assess an ankle injury. Complete the care.

Assess an ankle injury. Complete the care.

Months of symptoms. A new injury. An agent must review the evidence, update the chart, and arrange care that the patient can actually use.

Months of symptoms. A new injury. An agent must review the evidence, update the chart, and arrange care that the patient can actually use.

Months of symptoms. A new injury. An agent must review the evidence, update the chart, and arrange care that the patient can actually use.

Read the research

Performance and cost

Seven models on the same clinical assessment.

Mean task score / 100
0
20
40
60
GPT-6 Astra$0.3$1$3$10$30

Mean recorded API cost · USD, log scale

52.19/ 100 mean score
$2.11mean cost
The line connects models with the highest observed score at each cost. 8 scored attempts for GPT-6 Astra. Costs exclude grading and infrastructure.

Get in touch ↗