TL;DR
- The LTX 2.5 Fast family (text-to-video
ltx-2.5-fast/text-to-video+ image-to-videoltx-2.5-fast/image-to-video) bills per second: $0.043/s at 720p (a 6-second clip is $0.258) and $0.086/s at 1080P; both endpoints share pricing, and the audio switch is free. - Twelve enumerated durations (2–20 s); your bill is
duration× the per-second rate. $10 buys about 38 clips (720p, 6 s). - Both endpoints are live: the text-to-video page and the image-to-video page. New accounts get 200 free credits. Usage: the t2v guide.
Price table (text and image endpoints alike)
| Tier | Per second | 5 s clip | 6 s clip | 10 s clip | 100 clips (6 s) | $10 yields (6 s) |
|---|---|---|---|---|---|---|
| 720p (default) | $0.043 | $0.215 | $0.258 | $0.43 | $25.80 | 38 |
| 1080P | $0.086 | $0.43 | $0.516 | $0.86 | $51.60 | 19 |
HiAPI current prices, checked October 23, 2026. The generate_audio switch never changes the price; failed tasks refund.
Real outputs at both resolutions
Both clips are aerial shots, ~5 s, 16:9, with sound; the prompts differ (an alpine lake versus a futuristic city) — read the pair for resolution and sharpness only:
| 720p $0.215 (5 s) | 1080P $0.43 (5 s) |
|---|---|
| 720p alpine lake → | 1080P future city → |
Left: ltx-2.5-fast/text-to-video, 720p, 1280×720, ~5 s, audio, $0.215. Right: 1080P, 1920×1080, ~5 s, audio, $0.43. Different prompts — resolution and sharpness are the point.
What you can see:
- Both are drone views with ambient sound; 1080P's building edges and light trails read sharper.
- At 5 seconds, 1080P costs 2× 720p — pick per delivery platform, not by default.
Spending less
- Iterate at 720p: composition and motion read fine; upgrade keepers to 1080P.
- Audio is free:
generate_audio: falsegives clean footage at the same price — no penalty for scoring in post. - Match duration to content: one action fits 5–6 s;
durationtakes enumerated values only (2, 3, 4, 5, 6, 8, 10, 12, 14, 16, 18, 20). - Bulk at 720p: 100 six-second clips cost $25.80; vertical 9:16 is the same price.
Compared with related video models (HiAPI prices, per second)
| Model | 720p | 1080p | Notes |
|---|---|---|---|
grok-imagine-1.5-lite/text-to-video | $0.043 | $0.2 | Sound always on; duration any integer 1–15 |
ltx-2.5-fast/* (this family) | silent $0.043 / audio same | $0.086 | Free audio switch; 12 durations |
wan2.6-flash/image-to-video | silent $0.029 / audio $0.0575 | silent $0.043 / audio $0.086 | 5/10/15 s only; aspect follows image |
grok-imagine/text-to-video | $0.0214 | $0.057142 | 480p $0.0114 |
wan2.7-video/image-to-video | $0.1 | $0.167 | Input uses a media array |
Silent 720p is cheaper on Wan 2.6 Flash; for 1080P with controllable audio, LTX's $0.086 is a direct option.
Monthly budgets (720p, 6 s per clip)
| Clips / month | All 720p | All 1080P |
|---|---|---|
| 100 | $25.80 | $51.60 |
| 500 | $129.00 | $258.00 |
| 1,000 | $258.00 | $516.00 |
"720p drafts + 1080P finals" is the common mix: 400 six-second 720p drafts plus 100 six-second 1080P finals runs about $154.80 a month.
FAQ
- Yes, text and image endpoints cost the same — no input-image fee.
- Yes, muting saves nothing: only resolution and duration set the price.
- Yes, durations are 12 enumerated values (2–20); 7 or 9 returns a 400.
- Yes, $10 buys about 38 clips (720p, 6 s).
- Yes, failures are free; validation rejections never bill.
Next: take one prompt and run a 720p clip ($0.258 for 6 s); scene templates are in the t2v prompt guide.
