H3 Max
Cost calculator.
Your model. Your resolution. Your video budget, calculated.
1 clip × 10 seconds / H3 Max / 768p
Automatic uses Sep 15 for Max / Turbo and Sep 14 for Director, at 00:00 UTC in 2026. Dates come from their separate fal pages; UTC is our assumption because no cutoff timezone is published. Reference-to-video has no listed launch discount.
H3 Max cost at a glance.
Compare ordinary Max and Turbo at 768p. Long durations below are aggregate clip budgets.
Scroll sideways to compare Max and Turbo
| Generated output · 768p | Max launch / standard | Turbo launch / standard |
|---|---|---|
| 5 secondsOne short clip | $0.10$0.40 standard | $0.05$0.20 standard |
| 10 secondsOne clip | $0.20$0.80 standard | $0.10$0.40 standard |
| 15 secondsOne longest-length clip | $0.30$1.20 standard | $0.15$0.60 standard |
| 1 minute of output4 × 15-second clips | $1.20$4.80 standard | $0.60$2.40 standard |
| 15 minutes of output60 × 15-second clips | $18.00$72.00 standard | $9.00$36.00 standard |
These are finished-clip budgets, including native audio. Max and Turbo clips have no 60-second session minimum. One- and 15-minute totals require multiple requests; they do not represent a single continuous clip. Director has separate session billing.
The endpoint sets
the billing rules.
Ordinary Max and Turbo are the default clip workflow. Choose Director only when you need a live session.
seconds × rate × clip count- 01
Max & Turbo: pay for your clips
Choose 480p, 768p or 1080p and a 5-15 second duration. A 10-second 768p Max clip is $0.20 at launch; Turbo is $0.10. Neither uses a one-minute session minimum.
- 02
Reference: include the input budget
Reference-to-video uses standard output rates plus input charges above 4,096 tokens per request. Enter the combined token estimate for one clip; that allowance is applied again for each new request.
- 03
Director: apply a floor to each session
Director bills max(generated seconds, 60) × rate. Its public session cap is 15 minutes; prepaid credit can shorten it. A short session is $1.20 at launch or a calculated $4.80 at the standard rate.
Two price-change dates. Four different workflows.
fal’s Max / Turbo overview lists standard prices from September 15, 2026. Its Director page lists September 14. The reference endpoint has no listed launch offer. Automatic mode follows the selected endpoint; manual prices remain available.
See every fal H3 Max rate Compare other API providersFrequently asked questions.
A little clarity before you hit generate.
Which endpoint should I budget for my project?
For a product teaser or a short social post, start with an ordinary Max clip at the resolution you intend to publish. Compare Turbo using the same prompt, duration and resolution when evaluating a lower-cost option. If your brief depends on supplied reference media, select Reference and include its input tokens instead of applying the ordinary Max launch rate. For an interactive experience that changes direction during playback, estimate a Director session and apply its minimum separately to every new session. A collection of independent clips should use the clip count field; it should not inherit Director’s session minimum simply because the combined output lasts a minute. Save a launch-price estimate and a standard-price estimate for work that may run after the offer ends. Finally, allow a separate production budget for additional generations you decide to request: the calculator estimates generated output, while your final edit may use only a portion of it.
How much is a 10-second H3 Max clip?
At 768p, ordinary Max costs $0.20 at launch or $0.80 at standard price. Turbo costs $0.10 or $0.40. Reference-to-video costs $0.80 for output plus any billable input tokens. A separate Director session that produces 10 seconds still bills at least $1.20 at launch or a calculated $4.80 at standard price.
Why can I only enter 5-15 seconds for clips?
These clip endpoints accept short generations. For a minute or more of planned output, use the number of clips: four 15-second clips make 60 generated seconds. Director instead supports a live session estimate of up to 900 seconds, with a 60-second billing floor applied to every new session.
Does automatic pricing use the same cutoff for every model?
No. Max and Turbo switch to their standard rates on September 15, 2026; Director switches on September 14. Automatic mode assumes 00:00 UTC for those dates, since fal has not specified a timezone. Reference-to-video always uses its published standard output and input rates.
How do reference inputs affect the estimate?
Enter the combined input token estimate per clip. Each request includes 4,096 tokens; excess tokens cost $0.02 per 1,000, added to output duration times the resolution rate. The allowance is applied separately to each clip. Entering zero gives output cost only and does not represent a valid request without reference media.
Why are some estimates fractions of a cent?
fal publishes rates such as $0.00625 per second for Turbo at 480p during launch. Five seconds therefore costs $0.03125 before any provider rounding. The calculator preserves that precision. Estimates exclude taxes, recharge fees, retries and other services; the provider’s invoice determines the final charge.
Sources, not guesswork.
Provider pages checked Sep 9, 2026. Every comparison names the endpoint and resolution. Calculated totals are estimates for generated output, excluding taxes, retries and other services. We do not run an independent video-quality benchmark.
- fal · H3 Max & Turbo overview (opens in a new tab)
- fal · H3 Max text-to-video pricing (opens in a new tab)
- fal · H3 Max Turbo pricing (opens in a new tab)
- fal · H3 Max reference pricing (opens in a new tab)
- fal · H3 Max Director pricing (opens in a new tab)
- fal · Director API documentation (opens in a new tab)