LTX-2.5
LTX-2.5
LTX-2.5 comes in two variants: ltx-2-5-fast for speed and low cost, and ltx-2-5-pro for higher fidelity. Fast produces portrait and landscape video up to 4K; Pro tops out at 1080p.
LTX-2.5 improves every stage of generation on a stronger foundation: a Diffusion Video Decoder with industry-leading pixel quality, higher fidelity, native multi-shot scenes, and a better distilled model.
Native multi-shot means a single generation can produce multiple connected shots, holding character, scene, lighting, visual style, and voice consistent across cuts.
Supported endpoints
Both variants accept camera_motion, generate_audio: false for silent video, last_frame_uri on image-to-video, and automatic duration.
Support matrix
Resolutions, FPS values, and duration options supported by each variant.
Aspect ratios: 16:9 (landscape) and 9:16 (portrait).
ltx-2-5-pro generates at 720p and 1080p only. For 1440p or 4K, use ltx-2-5-fast.
Automatic duration
Send "duration": null and the model picks the length itself, from your prompt. Use it when the prompt describes the shot rather than a fixed slot — a one-line action stays short, a multi-shot sequence runs longer.
Available on ltx-2-5-fast and ltx-2-5-pro, on both API versions: text-to-video (async, sync) and image-to-video (async, sync).
Send duration with a value of null. The field is still required — leaving it out returns duration is required.
The result never exceeds the longest duration available for your resolution and frame rate in the table above.
Billing
You are charged for the video you actually get, at the usual per-second rate.
On a prepaid account, credits are held against the longest duration your resolution and frame rate allow, because the length is not known until generation finishes. The unused remainder is released when the job completes. Your balance therefore has to cover that maximum up front — a request that would have produced 6 seconds is still declined if you cannot cover 20 on ltx-2-5-fast, or 10 on ltx-2-5-pro. Postpaid accounts hold nothing and are invoiced for the generated length.
Automatic duration cannot be combined with last_frame_uri on image-to-video. A last frame fixes where the clip has to end, which requires a known length.