Skip to benchmark content
MotionBenchHosted by Baz Studio
Pilot datasetOne prompt · one trial · 33 routes · not the official Core rankingRead the limits
openai · O+C

GPT-5.6 Luna

gpt-5.6-luna
Pilot visual tierA
Human assessment
Precise axis/beam construction, clear semantic handoff, excellent value.

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

GPT-5.6 Luna — MotionBench