Pilot datasetOne prompt · one trial · 33 routes · not the official Core rankingRead the limits

openai · O+C

GPT 5.5

gpt-5.5

Pilot visual tierB

Human assessment

Distinct particle/contour construction and stable wordmark.

One diagnostic trial. This is not a durable model capability claim.