Added

• Add PixVerse commands for text-to-video, image-to-video, reference-guided video, transitions, and video extension.

Changed

• Support more concurrent local listeners with lower idle polling frequency and immediate event draining.

Fixed

• Honor the server's Retry-After delay when retrying local listener requests.

Removed

• Remove seedance-v1-lite from Seedance CLI contract metadata.
  Migration: Use seedance-v1-pro or another supported Seedance model.

Fixed

• Acknowledge each valid Local Callback Listener event before forwarding so local HTTP failures do not replay or block later events.

Removed

• Remove seedance-v1-lite from Seedance CLI contract metadata.
  Migration: Use seedance-v1-pro or another supported Seedance model.

Changed

• Describe and validate the documented Gemini Omni, Grok Imagine, and Topaz request fields.

Fixed

• Expose supported Wan Flash image-to-video duration values in generated command metadata; request defaults remain API-applied.

Added

• Add public commands for current Price Schedules and task reservation quotes.
• Add Flux 2 Max text-to-image and remix-image model options with their supported request constraints.
• Expose Kling O1 models and reference image, reference video, reference role, and audio-preservation fields in generated Kling commands.

Changed

• Send the CLI version during browser sign-in so the authorization page can identify an available update and link to its release.

Added

• Add Kling 2.6 motion-control model and conditional field requirements to generated Kling command help.

Added

• Add the Kling V3 Omni model, output resolution, sound, duration, aspect ratio, and frame constraints to generated Kling command help.
• Add seven Producer FUZZ versions to generated text-to-music model help and validation.
• Add CLI support for continuing completed Kling v2.5 Turbo videos.

Added

• Add Kling 2.6 model, mode, sound, duration, and frame constraints to generated Kling command help.
• Add the Midjourney extend-video command with task creation, polling, and public input help.
• Add Flux text-to-image and remix-image commands with generated request validation and task polling.
• Expose Veo 3.1 Lite model and input constraints in generated CLI help.
• Add Qwen Image text-to-image, remix-image, and edit-image commands with generated contract help.

Added

• Expose request-scoped reference audio entries for Fish Audio text-to-speech input files.
• Add lyric blending command flags and generated contract help.

Breaking

• Replace Grok Imagine image-to-video source_image_urls with scalar source_image_url.
  Migration: Pass --source-image-url URL or provide source_image_url in JSON input.

Added

• Add the synchronous runapi midjourney shorten-prompt command with generated input validation and help.
• Add synchronous text-to-speech commands for OpenAI TTS and Fish Audio.
• Expose model and text input help from the generated contract.
• Add Gemini Omni Flash Preview model selection and model-specific text-to-video help.
• Add the gemini-tts text-to-speech command with generated input help and validation.
• Add Seedream 5 Pro text-to-image and edit-image models to CLI help and validation.
• Expose advanced stem separation mode and supported stem names in CLI help and validation.
• Add the Producer text-to-music command with generated contract help and typed SDK wiring.

Changed

• Expose Seedream 5-Lite output_format values and default in CLI help and validation.
• Include minimum and maximum item counts for constrained array fields in generated CLI metadata and help.

Fixed

• Preserve API-provided error codes in JSON output and omit the code when the response does not provide one.
• Classify command exit and retry behavior by HTTP status or SDK error type instead of generating substitute codes.