You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Seedance 2.5 schemas default generate_audio to false (#490). Seedance25RequestSchema and Seedance25InputSchema (bytedance/seedance-2-5) now default input.generate_audio to false instead of true, and the modelInputSchemas["bytedance/seedance-2-5"] descriptor says so. This changes .schema.parse() output and apicity describe. KIE's own documented default is still true, and createTask still sends your request as written, so a request that omits the field still gets KIE's true. Migration: send generate_audio: false for a silent clip, and set generate_audio: true explicitly if you send parsed output and relied on the old default. The price does not depend on it.
Grok Imagine Video 1.5 preview schemas default nsfw_checker to false (#490). GrokVideo15PreviewRequestSchema and GrokVideo15PreviewInputSchema now default input.nsfw_checker to false instead of true, as KIE's docs and live model schema say, and the modelInputSchemas["grok-imagine-video-1-5-preview"] description says so. This changes .schema.parse() output and apicity describe. createTask still sends your request as written, so a request that omits the field gets KIE's own default. Migration: pass nsfw_checker: true to keep KIE's content filter on.
@apicity/machgen
reference_video_start_secs is now a list (#492). The field takes one start second per src_video_urls entry (number[]), as MachGen requires. The schema typed it as a single number, which MachGen rejects with HTTP 400 "Input should be a valid list", so no request that set it could succeed. Migration: pass an array with one entry per reference video, such as [2] to read a single reference video from its second second. Models that read every reference video from its start reject any non-zero entry.
New
machgen: price the T2V-only video models and check their aspect ratios (#493)
machgen: catalog video-to-video (R2V grids, UPSCALE) with reference-video pricing (#492)
machgen: catalog every image-to-video model with pricing and the billing endpoint (#489)
kie: add the Grok Imagine Video 1.5 preview 1080p tier (#486)
Updated
kie: default seedance-2-5 generate_audio and grok-imagine-video-1-5-preview nsfw_checker to false (#490)
cli: check fal queue submit input, not its body, in har report (#488)
har: match fal queue URLs whose endpoint id spans segments (#487)
