Back to rankings

Base44 AI Gateway Adds Video Generation Models

Base44View original changelog

Base44 extended its AI Gateway with video generation endpoints, letting backend functions create video from a prompt, animate a still image, or guide a clip with reference media. The gateway covers 11 models from Google, ByteDance, Kuaishou, MiniMax and xAI. Video jobs run asynchronously, and a dry_run parameter estimates credit cost up front.

Key Takeaways

  • Video generation is now available through the Base44 AI Gateway, so backend functions can create video without a separate provider account.
  • The gateway offers 11 models from Google, ByteDance, Kuaishou, MiniMax and xAI behind one integration.
  • Three modes are supported: prompt to video, image animation, and reference-guided clips.
  • Jobs are asynchronous: create a job, store its ID, then poll for completion.
  • A dry_run parameter estimates credit cost before a request is exposed to end users.
  • The release fills a gap that Base44 users had raised in feature requests for native AI video.

Video Joins the AI Gateway

Base44 added video generation endpoints to the AI Gateway, the SDK module that gives backend functions access to Base44's managed AI models. The AI Gateway previously covered text and other model calls through an OpenAI-compatible interface. Video is a new modality for it, so builders can add video features to their apps without signing up with separate model providers.

What the Endpoints Can Do

The gateway supports three video workflows. Developers can generate a video from a text prompt, animate a still image, or guide a clip using reference media. Base44 routes these requests across 11 models from five providers: Google, ByteDance, Kuaishou, MiniMax and xAI. This lets an app pick the model that fits its quality, style or cost needs behind one consistent integration.

Asynchronous Jobs and Cost Estimates

Video generation takes much longer than a typical text call, so the endpoints work asynchronously. A backend function creates a job, stores the returned job ID, and polls until the video is ready. Base44 also added a dry_run parameter that returns an estimate of the credits a request would consume without running it. This lets builders check costs before exposing video generation to end users.

Where to Learn More

The linked reference page, "Generate video with the AI Gateway", documents the endpoints, supported models, the polling pattern, frame images, input references and limits.