langchain-freeimagetovideo

Free image to video AI with no daily cap: start with no sign-up, keep going on a free account. Every video comes with sound and no watermark. This is the official LangChain integration for Free Image to Video: it gives any LangChain agent a tool that turns a prompt, or a photo plus a prompt, into a video clip.

pip install langchain-freeimagetovideo
export FREEIMAGETOVIDEO_API_KEY=sk-...   # create one at https://freeimagetovideo.ai/settings/apikeys

from langchain_freeimagetovideo import FreeImageToVideoGenerator

video = FreeImageToVideoGenerator()
video.invoke({"prompt": "a paper lantern drifting over a night river, slow push in"})
# -> "https://cdn.freeimagetovideo.ai/....mp4"

# image to video: the photo becomes the first frame
video.invoke({
    "prompt": "the camera slowly orbits the statue",
    "image_url": "https://example.com/statue.jpg",
    "duration": 6,
    "aspect_ratio": "9:16",
})

Hand it to any tool-calling agent:

from langchain.agents import create_agent
from langchain_freeimagetovideo import FreeImageToVideoGenerator, FreeImageToVideoStatus

agent = create_agent("openai:gpt-5", tools=[FreeImageToVideoGenerator(), FreeImageToVideoStatus()])
agent.invoke({"messages": [("user", "Make a 6-second vertical clip of a fox trotting through fresh snow")]})

Tools

The model is MiniMax H3. Video takes minutes: the generator waits up to timeout seconds (600 by default) and otherwise hands back the task id, so the agent can call the status tool later instead of blocking.

On the site the same model runs as image to video AI with sound and free text to video AI.

API key and plans

Create a key at freeimagetovideo.ai/settings/apikeys and pass it as api_key= or set FREEIMAGETOVIDEO_API_KEY. API runs need a paid plan: each run draws on the plan's full-speed allowance, then waits in the relaxed queue. See pricing.

Refusals come back to the agent as text rather than an exception: a missing or wrong key, no paid plan, not enough credits, too many runs at once, or a prompt the filter refused. The agent can reword or tell the user what to do instead of crashing the run.

Built on the freeimagetovideo Python SDK.

MIT licensed. Source: github.com/hanshs474/langchain-freeimagetovideo.