ZMIME

AURA 5B vs PAN 2

Compare documented capabilities and access. A dash means a comparable figure is unavailable.

Model specifications
SpecificationAURA 5BKyutaiPAN 2MBZUAI Institute of Foundation Models
TaskSpeech-driven portrait animationWorld simulation
Published capabilitiesA four-step causal model animates a portrait from streaming speech, rendering chunks from the available audio.Predicts what happens after an action in a starting scene.
Model size5B parameters—
AccessPublic student checkpoints and inference codePublic weights, full training code and inference code
Result—Simulated video conditioned on a scene and action
Operating limitsKyutai reports 9.7 fps end to end on one H100 at 720p for 25 Hz playback.—
SourcesModel weightsPublisher documentation
InputsImages, AudioText, Images
WeightsOpen weights · Snap Inc. Non-Commercial LicenseOpen weights · Apache-2.0
ReleasedNov 20, 2026Nov 10, 2026