Generate Video
Generate a video from a text prompt. Returns MP4.
Available models: google/veo-3.1-fast, google/veo-3.1-fast:replicate, google/gemini-omni-1.1-flash, bytedance/seedance-1-pro-fast, bytedance/seedance-1-pro-fast:fal, bytedance/seedance-2.0, bytedance/seedance-2.0-mini, bytedance/seedance-2.0-fast, alibaba/wan-2.6, alibaba/wan-2.6:replicate, alibaba/wan-2.6:fal, alibaba/wan-2.2-fast, alibaba/wan-2.2-fast:fal, alibaba/wan-2.7, alibaba/wan-3.0, alibaba/wan-3.0:fal, x-ai/grok-imagine-video, x-ai/grok-imagine-video:openrouter, x-ai/grok-imagine-video-1.5, x-ai/grok-imagine-video-1.5:fal, x-ai/grok-imagine-video-1.5-lite, x-ai/grok-imagine-video-1.5-lite:openrouter, bytedance/seedance-2.5, alibaba/happyhorse-1.1, heygen/heygen-video-1, minimax/minimax-h3, minimax/minimax-h3-max, minimax/minimax-h3-max-turbo, prunaai/p-video.
Use duration to set video length, aspectRatio for orientation, and audio where the selected model supports audio output.
You can pass reference images via the image parameter: image[0] is the start frame, and image[1] is the end frame for models with end_frame in video_capabilities.
Seedance 2.0, Seedance 2.5, Wan 3.0, and MiniMax H3 Max also accept reference_images, reference_videos, and reference_audios for guidance distinct from frame controls. Separate URLs with |; commas inside URLs are preserved.
Browse all available models and their video_capabilities at /image/models.
Request
- Base URL: https://gen.pollinations.ai
- URL: https://gen.pollinations.ai/video/{prompt}
- Auth: HTTP bearer
Path parameters
Text description of the video to generate
Query parameters
Model to use. See /image/models for the current canonical IDs and aliases.
Width in pixels. For images, exact pixels; flux-2-pro, flux-2-flex, and microsoft/mai-image-2.6-flash require multiples of 16 (MAI also needs at least 768 px per side and at most 2,359,296 total pixels). black-forest-labs/flux.1.1-pro requires 256–1440 px per side in multiples of 32 and at most 1.6 megapixels. For video models, used for aspect ratio; use resolution to select a resolution tier.
Height in pixels. For images, exact pixels; flux-2-pro, flux-2-flex, and microsoft/mai-image-2.6-flash require multiples of 16 (MAI also needs at least 768 px per side and at most 2,359,296 total pixels). black-forest-labs/flux.1.1-pro requires 256–1440 px per side in multiples of 32 and at most 1.6 megapixels. For video models, used for aspect ratio; use resolution to select a resolution tier.
Seed for reproducible results. Use -1 for random. Supported by: black-forest-labs/flux.1-schnell, black-forest-labs/flux.1.1-pro, tongyi-mai/z-image-turbo, bytedance/seedream-4.0, black-forest-labs/flux.2-klein-4b, bytedance/seedance-2.0. Other models ignore this parameter.
Safety features: comma-separated list of privacy, secrets, sexual, violence, shield, true, nsfw. true enables privacy,secrets; nsfw enables sexual,violence. Also accepted in the Pollinations-Safe header. Defaults to off; false and 0 are accepted as off.
Image quality level. Supported by gptimage, gptimage-large, gpt-image-2, and grok-imagine-image-2.0.
Reference image URL(s) for image editing or video generation. Separate multiple URLs with | or ,. Image models: Used for editing or style reference. Video models: image[0] is the starting frame; image[1] is the optional ending frame. See video_capabilities and max_reference_images on /image/models or /models for per-model support.
Video models only: public HTTP(S) image URLs for visual guidance, separate from first/last-frame controls. Separate multiple URLs with |; commas inside URLs are preserved. See video_capabilities on /image/models or /models for per-model support.
Video models only: public HTTP(S) video URLs for motion or style guidance. Separate multiple URLs with |; commas inside URLs are preserved. See video_capabilities on /image/models or /models for per-model support.
Video models only: public HTTP(S) audio URLs for audio-driven generation. Separate multiple URLs with |; commas inside URLs are preserved. See video_capabilities on /image/models or /models for per-model support.
Generate image with transparent background. Only supported by gptimage and gptimage-large.
Output resolution for image and video models that advertise resolutions in /models. The first advertised resolution is the default; requested tiers bill at their listed rate.
Video duration in seconds. Only applies to video models. Community models may omit this if the provider reports generated seconds; billing prefers reported duration and otherwise uses this value. google/gemini-omni-1.1-flash: 3-10s. veo: 4, 6, or 8s. seedance-pro: 2-10s. seedance-2.0: 4-15s; Mini: 4-10s; Fast: 4-5s. seedance-2.5: exactly 4s. minimax-h3: exactly 5s. minimax/minimax-h3-max and minimax/minimax-h3-max-turbo: 5, 10, or 15s. wan: 2-15s. wan-3.0: exactly 5s.
Aspect ratio. For bytedance/seedream-5.0-flash, overrides width/height; adaptive lets the provider choose. For video models, if not set, determined by explicit width/height; veo, google/gemini-omni-1.1-flash, seedance-2.5, minimax/minimax-h3-max, and minimax/minimax-h3-max-turbo otherwise default to 16:9. Most video models support 16:9 or 9:16; minimax-h3 supports only 16:9, while minimax/minimax-h3-max and minimax/minimax-h3-max-turbo also support 21:9, 4:3, 1:1, and 3:4.
Generate audio for the video. Only applies to video models. veo generates and bills audio only when set to true. seedance-2.0, seedance-2.0-mini, seedance-2.0-fast, seedance-2.5, and wan-3.0 generate audio unless set to false. wan2.6, wan-2.7, grok-imagine-video-1.5, google/gemini-omni-1.1-flash, minimax-h3, minimax/minimax-h3-max, and minimax/minimax-h3-max-turbo always generate audio regardless of this flag.
Response
Success - Returns the generated video
Changes
No recorded changes to this operation across all 1 revision of this API.