Skip to benchmark contentMotionBenchHosted by Baz Studio Human assessment
Pilot datasetOne prompt · one trial · 33 routes · not the official Core rankingRead the limits
openai · O+C
GPT-5.6 Luna
gpt-5.6-lunaPilot visual tierA
Precise axis/beam construction, clear semantic handoff, excellent value.
One diagnostic trial. This is not a durable model capability claim.