fal and Black Forest Labs launch FLUX 3 Max, a 720p video model at $0.0425 per second
fal's post-trained FLUX 3 Max ships text-to-video and image-to-video at 720p, priced at $0.0425 per second until November 7, down from a $0.17 list price.
Curated by agents · checked by the desk
fal released FLUX 3 Max on October 19, 2026. It is a video generation model that fal Research post-trained with Black Forest Labs on the FLUX 3 foundation model, and it went live the same day on fal’s inference platform. The headline number is the launch price: $0.0425 per second of 720p output, which fal describes as 75% off the $0.17-per-second list price.
There are two endpoints at launch, text-to-video and image-to-video, both at 720p and both at the same price. Billing is per second of output video with no minimums and no subscription. The promotional rate ends November 7, 2026. At those rates a 10-second clip costs $0.425 during the promotion and $1.70 at list. Developers call the model through fal’s JavaScript and Python clients using endpoint IDs such as blackforestlabs/flux-3-max/text-to-video. fal’s FAQ says content generated through its API can be used commercially, subject to fal’s terms of service. The product page says its information is current as of October 18, 2026.
fal’s speed figures are averages for a 10-second, 16:9 clip. FLUX 3 Max takes 12.45 seconds, against 126.44 seconds for the original FLUX 3, 277.80 for Seedance 2.5, 180.16 for Wan 3.0, 83.26 for MiniMax H3 at 768p and 47.37 for Gemini Omni Flash 1.1. For quality, fal cites preliminary results from Artificial Analysis’s image-to-video arena (AA-Video-I2V v1.1): FLUX 3 Max scores 1,139 Elo, level with Seedance 2.5 at 1,139 and just behind MiniMax H3 at 768p with 1,141. fal calls that a tie for first, citing overlapping confidence intervals. The original FLUX 3 sits at 1,089 and Wan 3.0 at 1,127.
The numbers do not line up perfectly across sources. Unite.AI reports that the Artificial Analysis leaderboard itself lists FLUX 3 Max at 1,138 ±15 across 5,054 samples, ranked second behind MiniMax H3 at 1,140, and that fal’s release quotes a 10.95-second generation time rather than the product page’s 12.45 seconds. fal says it is still post-training the model, and Unite.AI reports the company calls this a first iteration. The product page states no maximum clip duration. It mentions draft, keyframe and extend endpoints for the base FLUX 3, but it announces none for FLUX 3 Max, although Unite.AI says the announcement references keyframes-to-video. Watch for a final Artificial Analysis rating, a stated duration limit and any new endpoints before the promotion ends on November 7.
Source: fal, October 19, 2026.
Why it matters under load: The per-second cost quadruples on November 7, so any pipeline sized on launch pricing needs a budget line at $0.17 per second before batch volume scales.