P-Video is Pruna AI's multimodal video generation model, and its headline feature is a built-in draft mode that changes how you iterate.

Before committing to a full render, you can preview your 5-second 720p video in about 2.5 seconds. If the motion, framing, and timing look right, you run the full render,done in roughly 10 seconds. That feedback loop makes a real difference when you're testing prompts or dialing in a concept.

Beyond speed, P-Video handles text-to-video, image-to-video, and audio-to-video through a single endpoint. Native audio generation is built in with dialogue, sound effects, and background music, so you're not managing a separate audio pipeline. You can also import your own audio tracks and sync them to the generated visuals.

Pricing: $0.02/sec at 720p.

Try Pruna P-Video here.

Vidu Q3

Vidu Q3 comes from Shengshu Technology and is currently ranked #2 globally on Artificial Analysis benchmarks for AI video generation.

The standout capability here is native audio-video generation in a single pass (similar to LTX-2): dialogue, SFX, and background music generated simultaneously with the visuals, not added after. This produces tighter synchronization than post-processing approaches and removes an entire step from the workflow.

Clips go up to 16 seconds at up to 1080p, with support for multi-shot sequencing in a single generation. You can describe multiple camera angles and scene transitions in one prompt, and Q3 will handle the cuts. Text-to-video and image-to-video are both supported.

Pricing: $0.15/sec.

Try Vidu Q3 here.

New on Discord

Create something cool with the endpoints? Hop into the new #built-on-runpod channel on Discord to show it off!

Author profile: Brendan McKeag