Interactive serving

FlashDreams serving keeps a world-model session alive while inputs and outputs stream through the application loop.

Figure placeholder: persistent serving session

Replace this block with a diagram showing client input, serving session, pipeline cache, model step, streaming decoder, and output returning to the client.

Serving model

Live input

Application controls or sensor updates arrive continuously.

Warm session

Pipeline and cache state persist across updates.

Model step

Encoder, transformer, scheduler, and decoder advance the world.

Streamed output

Frames or latent output return without closing the session.

Reference integrations

• LingBot-World is the primary camera-control serving reference.

• NVIDIA OmniDreams shows closed-loop autonomous-vehicle simulation.

• Get Started provides the shortest command-level path for trying inference and serving side by side.

Serving implementation references

• Serving for serving API concepts and component mapping.

• Inference pipeline overview for runner/pipeline execution flow.

• integrations/lingbot/lingbot/webrtc for the WebRTC serving stack.