What is the difference between LTX-2.3 Fast and LTX-2.3 Pro?
LTX-2.3 comes in two tiers: Fast and Pro. Fast is optimised for speed and lower cost, making it ideal for drafts and iteration. Pro delivers higher fidelity output for final renders and commercial work — and is required for audio-to-video, retake, and extend.
What is the difference between LTX-2.3 Fast and LTX-2.3 Pro?
Both models belong to the LTX-2.3 family and use the same API endpoints. The difference is in output quality, cost, and which generation types each one supports.
LTX-2.3 Fast: for drafts and iteration
LTX-2.3 Fast is optimised for speed and cost. It's the right choice when you're experimenting with prompts, testing settings, or building workflows where you'll generate a high volume of videos before selecting the best. The output quality is strong — it's not a "low quality" tier — but it's tuned for throughput over maximum fidelity.
Fast also supports the longest durations: up to 20 seconds at 1080p/24fps, compared to 10 seconds for Pro.
LTX-2.3 Pro: for final output
LTX-2.3 Pro is optimised for visual quality. It produces sharper detail, more stable motion, and more accurate prompt adherence — making it the right choice for client-facing work, final renders, and commercial output.
Pro is also the only model that supports audio-to-video, retake, and extend. If your workflow involves any of those endpoints, Pro is required.
Side-by-side comparison
| LTX-2.3 Fast | LTX-2.3 Pro | |
|---|---|---|
| Best for | Drafts, iteration, high-volume workflows | Final renders, commercial output |
| Text-to-video | Yes | Yes |
| Image-to-video | Yes | Yes |
| Audio-to-video | No | Yes |
| Retake | No | Yes |
| Extend | No | Yes |
| Max duration at 1080p | 20 seconds | 10 seconds |
| Cost | Lower | Higher |
How to specify the model
Pass the model ID in your request body — use ltx-2-3-fast for Fast or ltx-2-3-pro for Pro:
{
"model": "ltx-2-3-fast" // or "ltx-2-3-pro"
}For current pricing by model and resolution, see What does the LTX API cost? For full parameter details, see the API reference.