any2i_create_task
ChatGPTCreate an image-generation task (credits charged) and return job_id immediately. Omit image_urls for text-to-image; pass it for image-to-image. The count must be within the model min_image_count~max_image_count range. prompt is required. After submission, poll any2i_get_task for the result.
any2i_get_options
ChatGPTAfter selecting a model and quality, return the credit cost, input-image count requirements, and crop specification. Omitted parameters use model defaults. Tell the user the credit cost and obtain confirmation before submitting. Image-generation credits depend only on quality, not ratio, image count, or prompt length.
any2i_get_task
ChatGPTGet image-generation task status. When job_status is success, image_urls contains the finished image URLs; when fail, job_reason contains the failure reason; other values mean queued or processing and require more polling.
any2i_list_models
ChatGPTList image-generation models and their supported quality, aspect ratios, and input-image count range. Include prompt-length limits and credit costs for each quality. The same tools cover text-to-image and image-to-image; whether a model accepts input images and how many is determined by its min_image_count / max_image_count.
create_upload_url
ChatGPTIssue a generic file upload URL. After receiving upload_url, use multipart/form-data POST to upload the file directly (form field: file). The upstream response returns the complete file URL in d. Uploaded files are stored temporarily and cleaned periodically; use the URL promptly.
get_pricing_url
ChatGPTReturn the URL of the public informational page that describes the available membership plans and what each plan includes. Use it only when the user asks which plans exist, or to explain why a feature or model is not available for the current account. Do not use it to steer the user toward a plan.
get_user_profile
ChatGPTFetch the current signed-in user profile, including basic information, credit balance, membership status, and expiration time.
i2v_create_task
ChatGPTCreate an image-to-video task from uploaded image URLs, charge the selected credits, and return task_id immediately. With kind=image, pass one image for image-to-video or two in [first frame, last frame] order for fixed endpoints, and crop both to the same concrete ratio. With kind=reference, pass 1-3 reference images in order and provide prompt. After submission, poll i2v_get_task for the result.
i2v_get_options
ChatGPTAfter selecting a model and parameters, return the credit cost and target image ratio and pixel dimensions. Omitted parameters use model defaults. Tell the user the credit cost and obtain confirmation before submitting. Cropping must be done locally (for example with ffmpeg or sips); this service does not accept image binaries. kind must match the value used with i2v_list_models.
i2v_get_task
ChatGPTGet image-to-video/reference-image video task status (both use the same query endpoint). When task_status is success, video_url is the finished video URL; when fail, task_reason contains the failure reason; other values mean queued or processing and require more polling.
i2v_list_models
ChatGPTList image-to-video models and their supported quality, duration, generation modes, and aspect ratios. Include movement amplitudes, image-count limits, prompt-length limits, and credit costs for each parameter combination. kind=image is single-image/fixed-endpoint video; kind=reference is reference-image video; the model lists are separate. Set end_frame=true to return only models supporting fixed first-to-last frames.
mv_create_story
ChatGPTSubmit audio and images to start story creation; credits are charged and task_id is returned immediately. Upload audio and images with create_upload_url before submitting. After submission, poll mv_get_story for the outline, cover, character art, storyline, and lyrics.
mv_get_options
ChatGPTReturn MV Agent quality, aspect ratios, audio-duration range, and prompt-length limit, plus credit costs for each stage (story creation per task, storyboard production by audio duration). Passing audio_duration_sec also returns an estimate for storyboard-production credits.
mv_get_production
ChatGPTGet stage 2 progress, including each segment image/video status and URL, plus the final composed MV result_url. Tasks are asynchronous; callers must poll.
mv_get_story
ChatGPTGet stage 1 progress and outputs: story outline, cover, character art, storyboard, and timestamped lyrics, with status for each item (doing/success/fail). Role, object, and scene assets are also returned in role_list, object_list, and scene_list. Tasks are asynchronous; callers must poll.
mv_list_models
ChatGPTList MV Agent models with their qualities, aspect ratios, audio-duration range, prompt-length limit, and per-stage credit costs. Pass models[].model to mv_get_options and mv_create_story to choose a model; when omitted, the first (default) model is used. The model is fixed once the task is created.
mv_regenerate_asset
ChatGPTRegenerate role or cover art and optionally describe the requested edit in natural language. Credits are charged; regeneration is asynchronous, so poll the corresponding status with mv_get_story.
mv_regenerate_segment
ChatGPTRegenerate a segment by index and optionally update its image and video prompts; credits are charged. After regeneration, call mv_submit_production(compose_only=true) to refresh the final MV.
mv_submit_production
ChatGPTFirst call after stage 1 confirmation: start storyboard production, charge by audio duration, and automatically compose the final MV after all segments finish. After regenerating a segment, call again with compose_only=true to refresh the final MV without charging credits. After submission, poll mv_get_production for progress.
mv_update_lyrics
ChatGPTUpdate timestamped lyrics. Call only after stage 1 completes and before storyboard production. No credits are charged.
mv_update_storyline
ChatGPTUpdate storyboard segments (index, start/end times, and description). Call only after stage 1 completes and before storyboard production. No credits are charged.
t2v_create_task
ChatGPTCreate a text-to-video task from a prompt, charge the selected credits, and return task_id immediately. No input images or pre-uploaded media are required. prompt is required, and ratio must be a concrete aspect ratio rather than free. After submission, poll t2v_get_task for the result.
t2v_get_options
ChatGPTAfter selecting a model and parameters, return the credit cost and final effective values. Omitted parameters use model defaults. Tell the user the credit cost and obtain confirmation before submitting. Text-to-video has no input images, so ratio must be a concrete aspect ratio rather than free.
t2v_get_task
ChatGPTGet text-to-video task status. When task_status is success, video_url is the finished video URL; when fail, task_reason contains the failure reason; other values mean queued or processing and require more polling.
t2v_list_models
ChatGPTList text-to-video models and their supported quality, duration, generation modes, and aspect ratios. Include movement amplitudes, prompt-length limits, and credit costs for each parameter combination. Text-to-video uses only a prompt; no input images, cropping, or upload step is required.