Building the next generation of multimodal AI experiences with fal in Microsoft Foundry
Microsoft and fal integrate Kling, LTX, and MiniMax video models into Microsoft Foundry to expand enterprise AI video generation options with public preview access.
Video generation workloads vary widely in requirements, from creative exploration to product motion concepts, necessitating specialized models rather than a one-size-fits-all approach. Microsoft and fal are addressing this by making Kling, LTX, and MiniMax video models available in public preview through Microsoft Foundry, offering text-to-video and image-to-video capabilities as first-party Azure services. The partnership aims to provide developers with multiple model choices to match specific creative, quality, and cost needs for their applications. By integrating these models into Foundry, Microsoft expands its platform to support diverse video generation workflows within a unified development environment.
fal, described as a full-stack generative media platform, supports over 1,500 models across image, video, audio, 3D, and generative worlds, providing inference, training, and infrastructure for production workloads. The collaboration with Microsoft enables developers to access a curated set of video models optimized for different creative and technical requirements, including speed, quality, and cost efficiency. The initial portfolio includes Kling v3 Pro, LTX-2.5 Fast, LTX-2.5 Pro, MiniMax H3 Max, and MiniMax H3 Max Turbo, each designed to address distinct use cases such as rapid iteration, high-fidelity output, or prompt adherence.
The integration emphasizes enterprise readiness, with models hosted in the US and Microsoft Foundry offering the first US-hosted Kling 3 endpoint. Pricing for video generation varies by model and resolution, allowing teams to balance creative demands with budget constraints. Microsoft plans to publish updated pricing on its Azure pricing page in the coming weeks, providing transparency for cost comparisons across LTX and MiniMax models.
Developers can explore the new video models in Microsoft Foundry to identify the best fit for their applications, with a session at Microsoft Ignite scheduled for November 19 to discuss the partnership and its implications. The models support both text-to-video and image-to-video generation, with features like synchronized audio, custom element support, and post-training optimizations to enhance quality and performance.