# EdenLive API

> EdenLive as a hosted realtime API. A proxy-guided generated video stream on main_video. $0.0021 per billable second.

Source: https://windflow.online/models/edenlive · Live video transformation · Windflow Labs

## What it is

EdenLive is a proxy-guided, 5-billion-parameter video world model from VAST AI. It uses source game frames to guide motion and a scene prompt plus an optional reference image to establish appearance. Windflow exposes it as a Preview session through the same video API and Playground as its other realtime models.

## Specifications

- **Architecture:** 5B autoregressive diffusion transformer with proxy-DINO control and LightTAE decoding
- **Resolution:** 1280 × 704
- **Playout:** 24 fps delivery target
- **Chunk:** 9 opening frames, then 12 frames per chunk
- **Control latency:** Chunk-based rendering; browser input-to-display latency has not been qualified
- **Inputs:** Source game footage or a live video track, a scene prompt, and an optional visual reference
- **Outputs:** A proxy-guided generated video stream on main_video
- **Delivery:** WebSocket (H.264) · WebRTC
- **Session protocol:** sdk.realtime.v2
- **Price:** $0.0021 per billable second

## Pricing

$0.0021 per billable session-second ($7.56/hr of billable time). Billing starts at the session's first generation_started event and stops when the session is released or ends. Queueing, worker startup, time before generation starts and the documented reconnect grace are not billed. After generation starts, paused or unwatched time remains billable; release the owner session to stop the meter.

## Use cases

- Restyling recorded game prototypes.
- Exploring scene appearance from live client-rendered video.
- Comparing proxy footage with a generated world side by side.

## Where it comes from

- [VAST-AI-Research/EdenLive on GitHub](https://github.com/VAST-AI-Research/EdenLive) — Apache-2.0; DINOv3 vision encoder has separate terms
- [VAST-AI/EdenLive weights](https://huggingface.co/VAST-AI/EdenLive)

## FAQ

### Can I control EdenLive with WASD?

Motion comes from the source video. Your application can render a game locally and send its video track; direct movement commands are not part of this Preview.

### Do I need a reference image?

A reference image is optional. The bundled scenes include a cover; other inputs use their first source frame when no image is supplied.

### Is browser latency qualified?

The native model loop has been benchmarked on H100, MI355X and RTX PRO 6000. The Windflow browser transport still requires its deployment media gate; the 24 fps delivery target is not an end-to-end latency measurement.

### How much does EdenLive cost on Windflow?

EdenLive costs $0.0021 per billable session-second ($7.56/hr of billable time). Billing starts at the session's first generation_started event and stops when the session is released or ends. Queueing, worker startup, time before generation starts and the documented reconnect grace are not billed. After generation starts, paused or unwatched time remains billable; release the owner session to stop the meter. There is no subscription.

### How do I get access to EdenLive?

Windflow is in private beta. Request access from the sign-up page, or ask for a live demo of EdenLive. Approved accounts start with $5 and four concurrent sessions.

### Is EdenLive always running?

No. Model capacity is brought up on demand rather than kept standing. Live readiness for each model is published at api.windflow.online/v1/status, and a live demo can be arranged on request.

## Related guides

- [Choosing a realtime video model: a decision guide for 12 hosted models](https://windflow.online/guides/choosing-a-realtime-video-model)
- [EdenLive compute measurements on H100, MI355X and RTX PRO 6000](https://windflow.online/guides/edenlive-gpu-compute-measurements)
- [How to use the EdenLive API](https://windflow.online/guides/edenlive-api)