Runway

• Runway Character Performance — Generate a character performance video (act_two) from a character image and a reference performance video or image.
• Runway Create Upload — Upload a temporary media file to Runway and receive a runway:// URI for use in reference_image_uris, prompt_image, video_uri, and other generation parameters.
• Runway Delete Task — Cancel or delete a Runway task by ID.
• Runway Get Organization — Get usage tier and credit balance for your Runway organization, including credit balance, tier limits, and current usage statistics.
• Runway Get Task — Retrieve a Runway generation task by ID. Returns status, output URLs, and metadata. Set wait_for_completion=true to block until the task finishes.
• Runway Image Reference Edit — Generate an image variation locked to 1-3 reference images using gen4_image_turbo. Upload reference images first (inline bytes or from an HTTPS URL) to obtain runway:// URIs.
• Runway Image To Video — Generate video from an input image URL or data URI. Default video_model is gen4_turbo with ratio 1280:720. Duration is model-specific (veo3.1: 4/6/8, gen3a_turbo: 5/10, veo3: 8). When wait_for_completion is false, poll the returned task id until status is SUCCEEDED.
• Runway Query Usage — Fetch credit usage data broken down by model and day for your Runway organization. Up to 90 days of data can be queried at a time.
• Runway Sound Effect — Generate sound effects from text using Runway (eleven_text_to_sound_v2).
• Runway Speech To Speech — Convert speech to another Runway preset voice in audio or video (eleven_multilingual_sts_v2).
• Runway Text To Image — Generate images from a text prompt using Runway. Default image_model is gen4_image with ratio 1024:1024. prompt_text is capped at 1000 UTF-16 code units. Ratio options are model-specific. gen4_image_turbo requires reference_image_uris (1-3). When wait_for_completion is false, poll the returned task id until status is SUCCEEDED. Models: gen4_image, gen4_image_turbo, gemini_2.5_flash.
• Runway Text To Speech — Generate speech from text using a Runway preset voice (eleven_multilingual_v2).
• Runway Text To Video — Generate video from a text prompt. Default video_model is veo3.1 with ratio 1280:720. prompt_text is capped at 1000 UTF-16 code units. veo3.1 and veo3.1_fast accept duration 4, 6, or 8 seconds; veo3 requires duration 8. When wait_for_completion is false, poll the returned task id until status is SUCCEEDED.
• Runway Upload Bytes — Upload image bytes to Runway and receive a runway:// URI for gen4_image_turbo reference_image_uris. For HTTPS sources, register the file from a public URL instead of inline bytes. Provide exactly one of: file_content_base64 (max 1 MB inline), file_url (image bytes), or local_path (absolute path on the user's computer). Use list_local_directory to discover paths, then pass local_path on upload tools. Install MCPBundles Desktop at https://www.mcpbundles.com/desktop if it is not connected yet.
• Runway Video Upscale — Upscale a video with Runway gen4_aleph (4x upscale, max 4096px per side). Provide video_uri and a short prompt_text.
• Runway Voice Dubbing — Dub audio to a target language using Runway (eleven_voice_dubbing).
• Runway Voice Isolation — Isolate voice from background audio using Runway (eleven_voice_isolation). Audio must be between 4.6 and 3600 seconds.
• Wait Runway Task — Poll a Runway generation task until it succeeds, fails, is canceled, or times out. Use after tools that return a task id with wait_for_completion=false.

Schemas and invoke metadata: Directory API