Skip to main content

​
October 2026

​
Changed

  • Asset Generator pricing reduced (from October 11, 2026)
    • The API now charges the same credits as the Creatify app.
    • Most models cost half what they did. For example, Kling 3.0 Pro 5s goes from 17 to 8.5 credits, and Veo 3.1 8s from 64 to 32.
    • Seedance models cost a quarter of what they did. For example, Seedance 2.5 720p 5s goes from 46 to 11.5 credits.
    • Models with their own API prices, such as Boreal-H3, are unchanged.
    • See Models and pricing for every model’s current price.

​
Added

  • Model pages for the Asset Generator: Models and pricing lists every model, grouped by family. Each model has its own page with its full price table, a ready-to-run request and every input parameter. These pages replace the price table on Create Asset Generator task.
  • Aurora 2.0 on the Aurora page: model_version: "aurora_v2" renders at 480p, 720p (default), 1080p or 2k for 0.4, 0.7, 1.4 or 2.8 credits per second, from up to 59.7 seconds of audio. The page now lists every request parameter. See Aurora 2.0.
  • Billing and retention for the Asset Generator, Aurora and Boreal: failed and rejected tasks are refunded automatically, and generated files stay available for at least 7 days. See Billing and retention.

​
Fixed

  • gpt-image-2 price: it was documented at a flat 1 credit, but the price depends on quality and resolution (0.5 to 9 credits). The model page lists every combination.

​
September 2026

​
Added

  • HTML Interactive Ad API: POST /api/html_interactive_ads/ turns a link into playable HTML ads, one self-contained file per style, ready to upload to an ad network. See HTML Interactive Ad.
  • Boreal API: POST /api/boreal/ generates a video from a prompt, a start image or an audio track. See Boreal.
  • Boreal-H3 on the Boreal API: model: "boreal-h3" on POST /api/boreal/ renders with Boreal-H3, Creatify’s post-trained H3 model. It takes an optional closing frame (end_image_url) or reference media (reference_image_urls, reference_video_urls, reference_audio_urls), and renders 5 to 15 seconds at 768p, 1088p or 2k. See Boreal-H3.
  • boreal option to the model_version field in URL to Video, AI Avatar v1, AI Avatar v2 and Custom Template endpoints
    • Cost: 10 credits per 30-second video.
    • AI Avatar v1: audio up to 60 seconds, text up to 800 characters, and no green_screen or transparent_background.
    • AI Avatar v2: each input_text up to 800 characters.
    • If omitted, the default remains standard.

​
Removed

  • Inspiration API (documented as AI Ad Template): GET /api/inspirations/, GET /api/inspirations/{id}/, GET /api/inspiration_jobs/, GET /api/inspiration_jobs/{id}/ and POST /api/inspiration_jobs/ are deprecated and no longer documented. For image and video generation from a template, use the Asset Generator API.

​
May 2026

​
Changed

  • Aurora pricing reduced
    • aurora_v1: now 1 credit per second (previously 20 credits per 15-second video).
    • aurora_v1_fast: now 0.5 credits per second (previously 10 credits per 15-second video).
    • Billing is now per-second (partial seconds round up to the next whole second), replacing the previous 15-second increments.

​
November 2025

​
Added

  • Asset Generator API: unified endpoints for image and video generation.
  • Inspiration API: endpoints for inspiration recreation.

​
October 2025

​
Added

  • aurora_v1_fast option to model_version field in aurora endpoints
    • Fast version of the aurora_v1 model that maintains high generation quality with faster processing.
    • Cost: 10 credits per 15-second video.
    • This option is now available alongside existing standard and aurora_v1 choices.

​
September 2025

​
Added

  • model_version field in both AI Avatar v1 and AI Avatar v2 endpoints
    • New optional parameter to specify which model powers the AI Avatar generation.
    • Choices:
      • standard – Default. 5 credits per 30-second video.
      • aurora_v1 – Higher-quality Aurora model. 20 credits per 15-second video.
    • If omitted, the default remains standard.
    • This field appears in both request and response payloads.

​
Why it matters

Developers can now explicitly select the underlying model for more control over cost and output quality.
Stay tuned for more improvements. Older releases will be documented here as new updates are rolled out.