accept_output_plan
ChatGPTAccept one exact output proposal revision within a USD cap. Use explicit Project/resource IDs; external chat has no browser page context. Show the inline card. Keep the accompanying reply to one short sentence; do not repeat card details, internal IDs, or tool logs. Quotes require the user to click Create or explicitly approve the displayed maximum before paid work. Use the conversation for creative changes; preserve all other settings.
add_source
ChatGPTAdd one URL, YouTube, text, Product Visual URL, or authorized ChatGPT file. Files support PDF, DOCX, TXT, Markdown, PNG, JPEG and WebP up to 20 MiB. File imports are request-only by default; set keep_as_project_asset:true only when the user wants it saved to the Project. Wait for ready before generation. No local paths or base64.
archive_content_loop
ChatGPTArchive a Content Loop. Use explicit Project/resource IDs; external chat has no browser page context.
archive_project
ChatGPTArchive a Project and its Loops. Use explicit Project/resource IDs; external chat has no browser page context.
cancel_generation
ChatGPTCancel undispatched Generation work. Use explicit Project/resource IDs; external chat has no browser page context.
cancel_output_plan
ChatGPTCancel remaining admissions and undispatched work. Use explicit Project/resource IDs; external chat has no browser page context.
create_collection
ChatGPTCreate a Source collection. Use explicit Project/resource IDs; external chat has no browser page context.
create_content_loop
ChatGPTCreate one recurring multi-Asset Content Loop with explicit trend, topic, or knowledge input and both per-run and monthly service-USD caps. A stored narration_script is deliberately reused unchanged on every run. Standard models for all five Video Assets accept narration_script for exact caller-authored text. Put Voice and Avatar IDs inside each speaker, never at the Asset root; options.presentation_mode remains authoritative. Two speakers require a Voice each and, in Avatar mode, an Avatar each. duration_seconds is an approximate target: complete shorter narration ends the video without synthetic trailing padding. Exact preserves accepted text, punctuation, attribution, and order with no truncation, but is not ASR pronunciation or OCR certification. Narrated video options offer caption_style (social_bold, social_highlight, clean), caption_font (inter, montserrat, oswald), and caption_position (top, center, bottom). captions controls burned-in text; every style uses a transparent overlay without a full-width dark box, and the VTT Artifact remains available. When get_capabilities with asset_type launch_video or ad_video and include_models:true lists autocontent-motion-design-v1, that instrumental model accepts integer duration_seconds 15–60. Launch defaults to 60 seconds, 16:9, 1080p/30 fps with optional 4k/60 fps. Ads default to 15 seconds, 9:16, 1080p/30 fps and also support 1:1 or 16:9; Ads do not support 4k. Omit Voice, Avatar, narration_script, and caption styling; captions defaults to false. model_options.music_direction accepts at most 500 characters. Every create or Loop occurrence gets an original instrumental. Ordinary edits retain the exact score; changed duration/music direction or edit-only refresh_music:true requires a newly quoted score. The refresh directive applies once. Preview is a maximum reservation; settlement uses actual production cost within it. When capabilities list autocontent-recorded-demo-v1 for product_demo_video, launch_video, or ad_video, provide options.website_url (publicly reachable HTTPS) with optional options.recording_access_id saved in the Project access form and instructions describing the journey. AI records real browser actions and composes footage, typography, optional narration, and an original arranged score. Duration is 15–120 seconds, 1080p, with 16:9, 9:16, or 1:1. model_options accepts narration (default true) and music_direction (500 characters). Browse is the default interaction_mode; demo requires demo_environment_confirmed:true attesting to a test environment without real purchases, messages, or destructive effects. Never put credentials in generation payloads or prompts. Saved access supports Basic or ordinary form login; SSO, MFA, CAPTCHA and private networks are unsupported. Interrupted interactions are not replayed automatically. Compatible visual edits retain verified footage, speech and score. Changed source settings or model_options.refresh_recording:true request fresh media. Preview states retained/new media and the spending cap. When discovery lists autocontent-narrated-video-v1 for short_video or explainer_video, use it for illustrated explainers with narration, rich artwork and explanatory animation. It accepts existing duration, aspect ratio, 720p/1080p, captions, exact Voice and narration_script settings, with faceless presentation and no music. Charon is its economical stock narrator; preserve a requested Voice and use the preview quote. A small amount of H3 animation is selected by the director; unused resources are not charged. The same model can be retained in edits and Loops.
create_custom_avatar
ChatGPTCreate a consent- and ownership-attested custom Avatar from a public MP4/QuickTime URL. Hosted MCP accepts no local path or base64 media.
create_custom_voice
ChatGPTCreate a consent- and ownership-attested custom Voice from a public MP3/WAV/M4A URL. Hosted MCP accepts no local path or base64 media.
create_project
ChatGPTCreate a Project from one public website URL. Analysis starts asynchronously.
delete_avatar
ChatGPTRevoke a Project-owned custom Avatar. Curated Avatars cannot be removed and video presentation is never downgraded.
delete_collection
ChatGPTDelete an empty Source collection. Use explicit Project/resource IDs; external chat has no browser page context.
delete_voice
ChatGPTRevoke a Project-owned custom Voice. Curated Voices cannot be removed and no substitute is selected.
edit_generation
ChatGPTCreate paid edits for 1–10 original Assets under max_cost_usd. Omitted fields retain accepted values; option objects and narration scripts replace whole objects, while narration_script:null returns to generated narration. Standard models for all five Video Assets accept narration_script for exact caller-authored text. Put Voice and Avatar IDs inside each speaker, never at the Asset root; options.presentation_mode remains authoritative. Two speakers require a Voice each and, in Avatar mode, an Avatar each. duration_seconds is an approximate target: complete shorter narration ends the video without synthetic trailing padding. Exact preserves accepted text, punctuation, attribution, and order with no truncation, but is not ASR pronunciation or OCR certification. Narrated video options offer caption_style (social_bold, social_highlight, clean), caption_font (inter, montserrat, oswald), and caption_position (top, center, bottom). captions controls burned-in text; every style uses a transparent overlay without a full-width dark box, and the VTT Artifact remains available. When get_capabilities with asset_type launch_video or ad_video and include_models:true lists autocontent-motion-design-v1, that instrumental model accepts integer duration_seconds 15–60. Launch defaults to 60 seconds, 16:9, 1080p/30 fps with optional 4k/60 fps. Ads default to 15 seconds, 9:16, 1080p/30 fps and also support 1:1 or 16:9; Ads do not support 4k. Omit Voice, Avatar, narration_script, and caption styling; captions defaults to false. model_options.music_direction accepts at most 500 characters. Every create or Loop occurrence gets an original instrumental. Ordinary edits retain the exact score; changed duration/music direction or edit-only refresh_music:true requires a newly quoted score. The refresh directive applies once. Preview is a maximum reservation; settlement uses actual production cost within it. When capabilities list autocontent-recorded-demo-v1 for product_demo_video, launch_video, or ad_video, provide options.website_url (publicly reachable HTTPS) with optional options.recording_access_id saved in the Project access form and instructions describing the journey. AI records real browser actions and composes footage, typography, optional narration, and an original arranged score. Duration is 15–120 seconds, 1080p, with 16:9, 9:16, or 1:1. model_options accepts narration (default true) and music_direction (500 characters). Browse is the default interaction_mode; demo requires demo_environment_confirmed:true attesting to a test environment without real purchases, messages, or destructive effects. Never put credentials in generation payloads or prompts. Saved access supports Basic or ordinary form login; SSO, MFA, CAPTCHA and private networks are unsupported. Interrupted interactions are not replayed automatically. Compatible visual edits retain verified footage, speech and score. Changed source settings or model_options.refresh_recording:true request fresh media. Preview states retained/new media and the spending cap. When discovery lists autocontent-narrated-video-v1 for short_video or explainer_video, use it for illustrated explainers with narration, rich artwork and explanatory animation. It accepts existing duration, aspect ratio, 720p/1080p, captions, exact Voice and narration_script settings, with faceless presentation and no music. Charon is its economical stock narrator; preserve a requested Voice and use the preview quote. A small amount of H3 animation is selected by the director; unused resources are not charged. The same model can be retained in edits and Loops. Show the inline card. Keep the accompanying reply to one short sentence; do not repeat card details, internal IDs, or tool logs. Quotes require the user to click Create or explicitly approve the displayed maximum before paid work. Use the conversation for creative changes; preserve all other settings.
feedback_asset
ChatGPTRecord useful or not_useful outcome feedback for one Asset. This does not regenerate or change billing.
feedback_content_loop_run
ChatGPTRecord Content Loop Run topic feedback. Use explicit Project/resource IDs; external chat has no browser page context.
generate
ChatGPTGenerate 1–10 distinct Assets. Selecting an Asset type is sufficient and instructions are optional. Asset IDs include article, lead_magnet, ebook, slides, infographic, quiz, podcast_episode, short_video, explainer_video, launch_video, product_demo_video, ad_video. attachment_source_ids may reference up to 20 ready, same-Project Sources for this request; Text Sources can ground any Asset, while Product Visuals can guide Lead Magnets, Product Demo Videos, and available Motion Design Ads and Launch Videos and do not replace text evidence. Import selected ChatGPT files using add_source and wait until they are ready first. This call can deduct up to max_cost_usd in service USD from promotion or funded PAYG balance; tax applies only to the separate funding purchase. Use preview_generation when the cap is unknown. A model is optional; model_options requires one. Standard models for all five Video Assets accept narration_script for exact caller-authored text. Put Voice and Avatar IDs inside each speaker, never at the Asset root; options.presentation_mode remains authoritative. Two speakers require a Voice each and, in Avatar mode, an Avatar each. duration_seconds is an approximate target: complete shorter narration ends the video without synthetic trailing padding. Exact preserves accepted text, punctuation, attribution, and order with no truncation, but is not ASR pronunciation or OCR certification. Narrated video options offer caption_style (social_bold, social_highlight, clean), caption_font (inter, montserrat, oswald), and caption_position (top, center, bottom). captions controls burned-in text; every style uses a transparent overlay without a full-width dark box, and the VTT Artifact remains available. When get_capabilities with asset_type launch_video or ad_video and include_models:true lists autocontent-motion-design-v1, that instrumental model accepts integer duration_seconds 15–60. Launch defaults to 60 seconds, 16:9, 1080p/30 fps with optional 4k/60 fps. Ads default to 15 seconds, 9:16, 1080p/30 fps and also support 1:1 or 16:9; Ads do not support 4k. Omit Voice, Avatar, narration_script, and caption styling; captions defaults to false. model_options.music_direction accepts at most 500 characters. Every create or Loop occurrence gets an original instrumental. Ordinary edits retain the exact score; changed duration/music direction or edit-only refresh_music:true requires a newly quoted score. The refresh directive applies once. Preview is a maximum reservation; settlement uses actual production cost within it. When capabilities list autocontent-recorded-demo-v1 for product_demo_video, launch_video, or ad_video, provide options.website_url (publicly reachable HTTPS) with optional options.recording_access_id saved in the Project access form and instructions describing the journey. AI records real browser actions and composes footage, typography, optional narration, and an original arranged score. Duration is 15–120 seconds, 1080p, with 16:9, 9:16, or 1:1. model_options accepts narration (default true) and music_direction (500 characters). Browse is the default interaction_mode; demo requires demo_environment_confirmed:true attesting to a test environment without real purchases, messages, or destructive effects. Never put credentials in generation payloads or prompts. Saved access supports Basic or ordinary form login; SSO, MFA, CAPTCHA and private networks are unsupported. Interrupted interactions are not replayed automatically. Compatible visual edits retain verified footage, speech and score. Changed source settings or model_options.refresh_recording:true request fresh media. Preview states retained/new media and the spending cap. When discovery lists autocontent-narrated-video-v1 for short_video or explainer_video, use it for illustrated explainers with narration, rich artwork and explanatory animation. It accepts existing duration, aspect ratio, 720p/1080p, captions, exact Voice and narration_script settings, with faceless presentation and no music. Charon is its economical stock narrator; preserve a requested Voice a…
get_account
ChatGPTGet the current Platform account. Use explicit Project/resource IDs; external chat has no browser page context.
get_asset
ChatGPTGet Asset detail and Artifacts. Use explicit Project/resource IDs; external chat has no browser page context.
get_avatar
ChatGPTGet canonical public Avatar state and exact compatibility tuples.
get_capabilities
ChatGPTDiscover the live catalog and strict option contracts, including supports_narration_script and narration_script_schema. Launch Asset IDs are article, lead_magnet, ebook, slides, infographic, quiz, podcast_episode, short_video, explainer_video, launch_video, product_demo_video, ad_video; the returned catalog is authoritative for extensions and model availability. Use asset_type launch_video or ad_video and include_models:true to discover Motion Design and its model-specific music-only option constraints.
get_content_loop
ChatGPTGet a Content Loop. Use explicit Project/resource IDs; external chat has no browser page context.
get_content_loop_run
ChatGPTGet a Content Loop Run. Use explicit Project/resource IDs; external chat has no browser page context.
get_generation
ChatGPTGet one Generation with resolved evidence, Asset states, artifacts, usage, and polling guidance. wait_seconds is bounded to 15 seconds.
get_output_plan
ChatGPTRead an output proposal and accepted Generations. Use explicit Project/resource IDs; external chat has no browser page context.
get_project
ChatGPTGet the editable inferred Project profile. wait_seconds may briefly wait up to 15 seconds for analysis.
get_source
ChatGPTGet Source ingestion status and revision. Use explicit Project/resource IDs; external chat has no browser page context.
get_usage
ChatGPTRead authoritative promotion/PAYG availability, active reservations, period usage, concurrency, and Project/Loop/provider-operation limits. This tool cannot fund or mutate billing.
get_voice
ChatGPTGet canonical public Voice state and exact compatibility tuples.
list_assets
ChatGPTList the Project Asset library. Use explicit Project/resource IDs; external chat has no browser page context.
list_collections
ChatGPTList Source collections. Use explicit Project/resource IDs; external chat has no browser page context.
list_content_loop_runs
ChatGPTList Content Loop Run history. Use explicit Project/resource IDs; external chat has no browser page context.
list_content_loops
ChatGPTList one Content Loop page including schedule, last outcome, current budget use, pause/block reason, and model warnings.
list_generations
ChatGPTList Generations. Use explicit Project/resource IDs; external chat has no browser page context.
list_projects
ChatGPTList one server-filtered Project page. Omission excludes archived Projects; request status archived explicitly.
list_sources
ChatGPTList one Project Source page with optional collection and bounded title-substring filters.
open_generation_panel
ChatGPTOpen the AutoContent workspace: project assets and sources, actual previews, generation or output-plan progress, and scoped edits. Omit selectors for the library. Optional project_id must own the selected resource. Opening never starts or retries paid work. Show the inline card. Keep the accompanying reply to one short sentence; do not repeat card details, internal IDs, or tool logs. Quotes require the user to click Create or explicitly approve the displayed maximum before paid work. Use the conversation for creative changes; preserve all other settings.
prepare_output_plan
ChatGPTValidate and price a bounded ordered set of outputs without inference. Use explicit Project/resource IDs; external chat has no browser page context. Show the inline card. Keep the accompanying reply to one short sentence; do not repeat card details, internal IDs, or tool logs. Quotes require the user to click Create or explicitly approve the displayed maximum before paid work. Use the conversation for creative changes; preserve all other settings.
preview_generation
ChatGPTValidate a canonical Generation draft and return its current maximum service-USD reservation without provider work or usage. attachment_source_ids may reference up to 20 ready, same-Project Sources for this request. Text Sources can ground any Asset; Product Visuals can guide Lead Magnets, Product Demo Videos, and available Motion Design Ads and Launch Videos and do not replace text evidence. Import user-selected ChatGPT files through add_source, then wait for Source readiness. Standard models for all five Video Assets accept narration_script for exact caller-authored text. Put Voice and Avatar IDs inside each speaker, never at the Asset root; options.presentation_mode remains authoritative. Two speakers require a Voice each and, in Avatar mode, an Avatar each. duration_seconds is an approximate target: complete shorter narration ends the video without synthetic trailing padding. Exact preserves accepted text, punctuation, attribution, and order with no truncation, but is not ASR pronunciation or OCR certification. Narrated video options offer caption_style (social_bold, social_highlight, clean), caption_font (inter, montserrat, oswald), and caption_position (top, center, bottom). captions controls burned-in text; every style uses a transparent overlay without a full-width dark box, and the VTT Artifact remains available. When get_capabilities with asset_type launch_video or ad_video and include_models:true lists autocontent-motion-design-v1, that instrumental model accepts integer duration_seconds 15–60. Launch defaults to 60 seconds, 16:9, 1080p/30 fps with optional 4k/60 fps. Ads default to 15 seconds, 9:16, 1080p/30 fps and also support 1:1 or 16:9; Ads do not support 4k. Omit Voice, Avatar, narration_script, and caption styling; captions defaults to false. model_options.music_direction accepts at most 500 characters. Every create or Loop occurrence gets an original instrumental. Ordinary edits retain the exact score; changed duration/music direction or edit-only refresh_music:true requires a newly quoted score. The refresh directive applies once. Preview is a maximum reservation; settlement uses actual production cost within it. When capabilities list autocontent-recorded-demo-v1 for product_demo_video, launch_video, or ad_video, provide options.website_url (publicly reachable HTTPS) with optional options.recording_access_id saved in the Project access form and instructions describing the journey. AI records real browser actions and composes footage, typography, optional narration, and an original arranged score. Duration is 15–120 seconds, 1080p, with 16:9, 9:16, or 1:1. model_options accepts narration (default true) and music_direction (500 characters). Browse is the default interaction_mode; demo requires demo_environment_confirmed:true attesting to a test environment without real purchases, messages, or destructive effects. Never put credentials in generation payloads or prompts. Saved access supports Basic or ordinary form login; SSO, MFA, CAPTCHA and private networks are unsupported. Interrupted interactions are not replayed automatically. Compatible visual edits retain verified footage, speech and score. Changed source settings or model_options.refresh_recording:true request fresh media. Preview states retained/new media and the spending cap. When discovery lists autocontent-narrated-video-v1 for short_video or explainer_video, use it for illustrated explainers with narration, rich artwork and explanatory animation. It accepts existing duration, aspect ratio, 720p/1080p, captions, exact Voice and narration_script settings, with faceless presentation and no music. Charon is its economical stock narrator; preserve a requested Voice and use the preview quote. A small amount of H3 animation is selected by the director; unused resources are not charged. The same model can be retained in edits and Loops. Show the inline card. Keep the accompanying reply to one short sentence; do not repeat card details, internal IDs, or tool logs. Quotes require the user to click Create or explicitly approve the displayed maximum b…
preview_generation_edit
ChatGPTValidate and quote edits for 1–10 unique original Assets without dispatching work or changing the Generation input/context. Omit narration_script to retain it, provide a complete object to add/replace it, or use null to return to generated narration. Standard models for all five Video Assets accept narration_script for exact caller-authored text. Put Voice and Avatar IDs inside each speaker, never at the Asset root; options.presentation_mode remains authoritative. Two speakers require a Voice each and, in Avatar mode, an Avatar each. duration_seconds is an approximate target: complete shorter narration ends the video without synthetic trailing padding. Exact preserves accepted text, punctuation, attribution, and order with no truncation, but is not ASR pronunciation or OCR certification. Narrated video options offer caption_style (social_bold, social_highlight, clean), caption_font (inter, montserrat, oswald), and caption_position (top, center, bottom). captions controls burned-in text; every style uses a transparent overlay without a full-width dark box, and the VTT Artifact remains available. When get_capabilities with asset_type launch_video or ad_video and include_models:true lists autocontent-motion-design-v1, that instrumental model accepts integer duration_seconds 15–60. Launch defaults to 60 seconds, 16:9, 1080p/30 fps with optional 4k/60 fps. Ads default to 15 seconds, 9:16, 1080p/30 fps and also support 1:1 or 16:9; Ads do not support 4k. Omit Voice, Avatar, narration_script, and caption styling; captions defaults to false. model_options.music_direction accepts at most 500 characters. Every create or Loop occurrence gets an original instrumental. Ordinary edits retain the exact score; changed duration/music direction or edit-only refresh_music:true requires a newly quoted score. The refresh directive applies once. Preview is a maximum reservation; settlement uses actual production cost within it. When capabilities list autocontent-recorded-demo-v1 for product_demo_video, launch_video, or ad_video, provide options.website_url (publicly reachable HTTPS) with optional options.recording_access_id saved in the Project access form and instructions describing the journey. AI records real browser actions and composes footage, typography, optional narration, and an original arranged score. Duration is 15–120 seconds, 1080p, with 16:9, 9:16, or 1:1. model_options accepts narration (default true) and music_direction (500 characters). Browse is the default interaction_mode; demo requires demo_environment_confirmed:true attesting to a test environment without real purchases, messages, or destructive effects. Never put credentials in generation payloads or prompts. Saved access supports Basic or ordinary form login; SSO, MFA, CAPTCHA and private networks are unsupported. Interrupted interactions are not replayed automatically. Compatible visual edits retain verified footage, speech and score. Changed source settings or model_options.refresh_recording:true request fresh media. Preview states retained/new media and the spending cap. When discovery lists autocontent-narrated-video-v1 for short_video or explainer_video, use it for illustrated explainers with narration, rich artwork and explanatory animation. It accepts existing duration, aspect ratio, 720p/1080p, captions, exact Voice and narration_script settings, with faceless presentation and no music. Charon is its economical stock narrator; preserve a requested Voice and use the preview quote. A small amount of H3 animation is selected by the director; unused resources are not charged. The same model can be retained in edits and Loops. Show the inline card. Keep the accompanying reply to one short sentence; do not repeat card details, internal IDs, or tool logs. Quotes require the user to click Create or explicitly approve the displayed maximum before paid work. Use the conversation for creative changes; preserve all other settings.
read_source_chunk
ChatGPTRead an exact published Project Source chunk without inference. Use explicit Project/resource IDs; external chat has no browser page context.
refresh_project
ChatGPTRefresh Project website analysis. Use explicit Project/resource IDs; external chat has no browser page context.
refresh_source
ChatGPTRefresh a remote Source. Use explicit Project/resource IDs; external chat has no browser page context.
remove_project_logo
ChatGPTRemove the current Project logo. Previously accepted Generations retain their frozen logo revision.
remove_source
ChatGPTRemove a Source from future Generations. Use explicit Project/resource IDs; external chat has no browser page context.
run_content_loop
ChatGPTStart one immediate Loop Run under the caller max_cost_usd and all stored Loop budget limits. It never changes the schedule or resumes a paused Loop.
search_knowledge
ChatGPTSearch matching indexed Project Source text without inference. Use explicit Project/resource IDs; external chat has no browser page context.
set_project_logo
ChatGPTCreate or replace the one Project logo from a public PNG/JPEG/WebP URL, freezing a new immutable revision.
update_collection
ChatGPTRename a Source collection. Use explicit Project/resource IDs; external chat has no browser page context.
update_content_loop
ChatGPTUpdate a Loop configuration or set status active/paused. Replacing assets replaces their complete narration_script configuration. Every call requires a fresh idempotency_key. Standard models for all five Video Assets accept narration_script for exact caller-authored text. Put Voice and Avatar IDs inside each speaker, never at the Asset root; options.presentation_mode remains authoritative. Two speakers require a Voice each and, in Avatar mode, an Avatar each. duration_seconds is an approximate target: complete shorter narration ends the video without synthetic trailing padding. Exact preserves accepted text, punctuation, attribution, and order with no truncation, but is not ASR pronunciation or OCR certification. Narrated video options offer caption_style (social_bold, social_highlight, clean), caption_font (inter, montserrat, oswald), and caption_position (top, center, bottom). captions controls burned-in text; every style uses a transparent overlay without a full-width dark box, and the VTT Artifact remains available. When get_capabilities with asset_type launch_video or ad_video and include_models:true lists autocontent-motion-design-v1, that instrumental model accepts integer duration_seconds 15–60. Launch defaults to 60 seconds, 16:9, 1080p/30 fps with optional 4k/60 fps. Ads default to 15 seconds, 9:16, 1080p/30 fps and also support 1:1 or 16:9; Ads do not support 4k. Omit Voice, Avatar, narration_script, and caption styling; captions defaults to false. model_options.music_direction accepts at most 500 characters. Every create or Loop occurrence gets an original instrumental. Ordinary edits retain the exact score; changed duration/music direction or edit-only refresh_music:true requires a newly quoted score. The refresh directive applies once. Preview is a maximum reservation; settlement uses actual production cost within it. When capabilities list autocontent-recorded-demo-v1 for product_demo_video, launch_video, or ad_video, provide options.website_url (publicly reachable HTTPS) with optional options.recording_access_id saved in the Project access form and instructions describing the journey. AI records real browser actions and composes footage, typography, optional narration, and an original arranged score. Duration is 15–120 seconds, 1080p, with 16:9, 9:16, or 1:1. model_options accepts narration (default true) and music_direction (500 characters). Browse is the default interaction_mode; demo requires demo_environment_confirmed:true attesting to a test environment without real purchases, messages, or destructive effects. Never put credentials in generation payloads or prompts. Saved access supports Basic or ordinary form login; SSO, MFA, CAPTCHA and private networks are unsupported. Interrupted interactions are not replayed automatically. Compatible visual edits retain verified footage, speech and score. Changed source settings or model_options.refresh_recording:true request fresh media. Preview states retained/new media and the spending cap. When discovery lists autocontent-narrated-video-v1 for short_video or explainer_video, use it for illustrated explainers with narration, rich artwork and explanatory animation. It accepts existing duration, aspect ratio, 720p/1080p, captions, exact Voice and narration_script settings, with faceless presentation and no music. Charon is its economical stock narrator; preserve a requested Voice and use the preview quote. A small amount of H3 animation is selected by the director; unused resources are not charged. The same model can be retained in edits and Loops.
update_project
ChatGPTUpdate only canonical Project profile fields; confirmed:true can atomically confirm a reviewed profile.