Veo (via Fal.ai)¶

Google's Veo3 and Veo3.1 models hosted on Fal.ai.

Quick Example¶

from tarash.tarash_gateway import generate_video
from tarash.tarash_gateway.models import VideoGenerationConfig, VideoGenerationRequest

config = VideoGenerationConfig(
    provider="fal",
    model="fal-ai/veo3.1",
    api_key="YOUR_FAL_KEY",
)

request = VideoGenerationRequest(
    prompt="A hummingbird hovering over a flower, slow motion",
    duration_seconds=6,
    aspect_ratio="16:9",
    generate_audio=True,
    seed=42,
)

response = generate_video(config, request)
print(response.video)

Supported Models¶

Model prefix	Duration	Image-to-Video	Notes
`fal-ai/veo3`	`4s`, `6s`, `8s`	✅	Audio; first/last frame; extend-video
`fal-ai/veo3.1`	`4s`, `6s`, `7s`, `8s`	✅	Fast variant via `/fast`; extend via `/fast/extend-video`

Parameters¶

Parameter	Required	Notes
`prompt`	✅
`duration_seconds`	—	`4`, `6`, `8` for veo3; `4`, `6`, `7`, `8` for veo3.1
`aspect_ratio`	—	Passed through directly
`resolution`	—	e.g. `"720p"`, `"1080p"`
`image_list` (reference)	—	Image-to-video
`image_list` (first_frame)	—	First frame pinning
`image_list` (last_frame)	—	Last frame pinning
`video`	—	Video extend
`seed`	—	Reproducibility
`negative_prompt`	—	Elements to avoid in the output
`generate_audio`	—	Generate audio track with the video

Video Extend¶

# Extend an existing video with veo3.1 fast
config = VideoGenerationConfig(
    provider="fal",
    model="fal-ai/veo3.1/fast/extend-video",
)

request = VideoGenerationRequest(
    prompt="Continue the scene in winter",
    duration_seconds=7,
    video="https://example.com/source.mp4",
)

Provider-Specific Notes¶

Veo3.1 vs veo3: Use fal-ai/veo3.1 for the latest model. Both use the same field mappers, so switching is transparent. fal-ai/veo3.1/fast also matches the fal-ai/veo3.1 prefix.

Extend-video constraints: The fal-ai/veo3.1/fast/extend-video endpoint has stricter limits: - Duration: 7s only - Resolution: 720p only - Requires both prompt and a video input