Test 01 · Paired benchmark
Garden Post Office
Jev and OpenAI Decisions sort identical letters into animated mailboxes. Adjust arrival rate, deadline, context and concurrency; compare latency, correctness and cost.
How the comparison worksWatch the demo
Meet Maple, the plush raccoon. In this recording, typed messages drive expression choices through OpenAI Decisions. Press Play to watch the reactions.
This page plays a recording without API calls or credentials. Run the app locally to interact with Maple or benchmark the two models.
Test 01 · Paired benchmark
Jev and OpenAI Decisions sort identical letters into animated mailboxes. Adjust arrival rate, deadline, context and concurrency; compare latency, correctness and cost.
How the comparison worksTest 02 · Voice and expressions
Speak with Maple, try typed messages or run the scripted emotion benchmark. OpenAI handles the voice and expression decisions in this test.
Run Maple locally