tongflow-api-runway

July 25, 2026 · View on GitHub

A TongFlow API plugin for the Runway Dev unified API (dev.runwayml.com, launched 2026-07-17). Runway Dev exposes Runway's own frontier models plus hosted third-party models behind a single API, selected per node via a model picker.

Setup

Add RUNWAYML_API_SECRET in TongFlow Settings (create one at dev.runwayml.com).

EnvDefaultPurpose
RUNWAYML_API_SECRETRequired. Runway Dev API secret.
RUNWAY_API_VERSION2024-11-06X-Runway-Version header.
RUNWAY_RATIOnode sizeForce an output ratio (e.g. 1280:720).
RUNWAY_TTS_VOICEDefault ElevenLabs voice id.
RUNWAY_POLL_TIMEOUT_S600Max seconds to wait for a generation.

Slots & models

ABI slotEndpointModels (first = default)
text-gen-video/text_to_videogen4.5, gen4_turbo, seedance2, seedance2_fast, veo3.1, veo3, happyhorse_1_0
image-gen-video/image_to_videogen4.5, gen4_turbo, seedance2, veo3.1, veo3.1_fast
image-image-gen-video/image_to_video (keyframes)gen4_turbo, gen4.5
images-gen-video/image_to_video (multi-ref)gen4.5, seedance2
video-edit/video_to_video (Aleph)aleph2
video-image-gen-video-move/character_performance (Act-Two)act_two
image-gen/text_to_imagegpt_image_2, gen4_image_turbo, seedream5_pro, gemini_image3_pro, seedream5_lite
image-edit/text_to_image (reference)gpt_image_2, seedream5_pro, gemini_image3_pro, gen4_image_turbo
text-gen-speech-preset/text_to_speechelevenlabs_tts, seed_audio

Act-Two (video-image-gen-video-move): the driving performance video is the reference, the target subject image is the character.

Runway is video-first; it does not serve text generation, transcription, music, or 3D. Model ids and the API version header rotate — verify against the live dev portal before relying on a specific id.

Licence

See the TongFlow repository.