SSyncDev

AURA-5B

SyncDev model directory · Profile updated Nov 21, 2026

Compare with another model

AURA-5B is Kyutai's larger streaming talking-head model: portrait plus speech to 720p, 25 fps video in four steps per chunk, at about 9.7 fps on one H100.

Key specifications

Specification score
20.0
License
Snap Inc. Non-Commercial License

The specification score summarizes recorded features and limits. It is not a benchmark result.

Pricing conditions: Open weights, non-commercial; no hosted API price.

Inputs and outputs

Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.

Accepts

  • Image input
  • Audio input
  • Text input

Produces

  • Video output

Capabilities

VisionAudioOpen weights

What is AURA-5B?

AURA-5B trades real time for resolution. It is the bigger sibling in Kyutai's AURA release of 18 November 2026, producing 720p talking-head video at 25 fps from one portrait and a speech track, and it does so in a streaming, chunk-by-chunk way. What it cannot do on a single H100 is keep up with its own playback rate.

Kyutai's own description contains that tension. Its headline promises generation "in real time"; its list of limitations says plainly that 9.7 fps end-to-end sits below 25 Hz playback. The second statement is the accurate one. If a live avatar is the goal, Kyutai points to AURA-1.3B instead.

Where the speed goes

  • Paper, optimised pipeline: 9.7 fps end-to-end at 720p, with the diffusion transformer at 19.7 fps and the VAE at 22.7 fps.
  • Reference code with --compile: about 8 fps on one H100.
  • Versus its teacher: 17 times faster end-to-end and 30 times faster on the transformer alone.

The VAE is nearly as slow as the transformer here, so optimising one without the other buys little. For offline rendering, a podcast clip or a batch of explainer videos, 8–10 fps is perfectly workable.

Built on someone else's base

Kyutai did not train this one from its own foundation. The teacher is Snap's 50-step bidirectional TalkVerse-S2V-5B, itself built on Wan2.2-TI2V-5B, and the student keeps that lineage. Three changes make it causal: block-wise causal attention over four-latent-frame chunks (about 0.64 s each), no motion-frame conditioning, and mixed diffusion timesteps per chunk. Audio comes from a wav2vec2 large XLSR-53 English encoder, not Kyutai's Mimi codec.

Training ran on 64 H100s in two LoRA stages (rank 128): ODE distillation, then distribution matching distillation. The final file, stage2_dmd/step-5000_merged.safetensors, is a standalone fp32 transformer of 5.7B parameters and 22.8 GB. LoRA-only and stage-1 checkpoints are there for researchers.

Because every file derives from TalkVerse, the weights inherit the Snap Inc. Non-Commercial License. The Wan2.2 VAE and T5 encoder you fetch separately remain Apache-2.0.

Quality, as reported

On the HALLO3 test split Kyutai reports FID 56.9, FVD 195.5 and Sync-C 4.30. In a user study of 400 pairwise comparisons from 27 raters, the model beat Hallo3 on visual quality, lip-sync, reference fidelity and naturalness, and matched or exceeded its own teacher. Yet Kyutai concedes that lip-sync trails the strongest multi-step systems, with Hallo3 at 5.06 and Wan-S2V at 5.42 on Sync-C.

A short test plan

Render the same 30-second English clip with both AURA sizes and with any multi-step baseline you already use. Look at mouth shapes on plosives, at teeth and at identity drift after a minute. Then time the run on your hardware, not ours or Kyutai's. Keep to four sampling steps; one-step sampling degrades sharply. Other languages and non-speech audio are untested.

The same consent and disclosure rules apply as for any face-animation tool. Kyutai's other models are on its provider page, and comparable systems are listed under audio-driven video, including Wan Animate 2.

How to evaluate AURA-5B

Compare the model on the work you need to complete.

Tasks and inputs

AURA-5B is listed for Image-to-video, Multimodal, Audio-Driven Video. It accepts image, audio, text and produces video. Use the publisher’s input format and preprocessing for your chosen deployment.

Capacity and cost

Choose duration, resolution and frame rate together. Compare the cost of an accepted clip, including repeat generations.

Choosing a deployment

The weights are downloadable under Snap Inc. Non-Commercial License. Check runtime support and size the complete model and working context for your expected concurrency.

Frequently asked questions

No. Kyutai reports 9.7 fps end-to-end at 720p on one H100 (about 8 fps with the reference code), below its 25 fps playback; AURA-1.3B is the real-time option.

These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.

kyutai

AURA-1.3B

AURA-1.3B is Kyutai's real-time streaming talking-head model: a portrait plus live speech becomes 576p, 16 fps video, chunk by chunk, on one GPU.

— context

Alibaba

Wan-Dancer-14B

Wan-Dancer-14B is Alibaba Wan's open music-to-dance video model, planning a whole track before refining it into minute-scale coherent dance.

— context

skywork

SkyReels-V3-A2V-19B

SkyReels-V3-A2V-19B is Skywork's open 720P talking-avatar model that animates a reference image from up to 200 seconds of driving audio.

— context

Meituan LongCat

LongCat Video Avatar 1.5

LongCat Video Avatar 1.5 generates audio-driven character video from speech, scene instructions and optional image references.

— context