Announced November 5, 2026, the fal partnership adds a curated set of text-to-video and image-to-video models to Microsoft Foundry, with Foundry billed as the first provider to host a US-based Kling 3 endpoint.
2026-11-07 · by Moe Ameen
On November 5, 2026, Microsoft announced that it is bringing five AI video models served by fal into Microsoft Foundry, its platform for building and running AI applications on Azure. The models are rolling out in public preview: Kling v3 Pro, LTX-2.5 Fast, LTX-2.5 Pro, MiniMax H3 Max, and MiniMax H3 Max Turbo. fal, which Microsoft describes as "the full stack for generative media" with APIs for more than 1,500 models across image, video, audio, 3D, and generative worlds, is supplying a curated selection of video models rather than its entire catalog; Microsoft says more models are planned.
The five span the usual generation modes. Most handle both text-to-video and image-to-video, and audio is built in across the set rather than bolted on afterward. Microsoft describes Kling v3 Pro as combining cinematic visuals and fluid motion with native audio generation and custom-element support; it runs on a US Data Zone deployment, and Microsoft calls Foundry the first provider to offer a US-hosted Kling 3 endpoint. LTX-2.5 ships as a Fast variant tuned for rapid iteration and a Pro variant tuned for high-fidelity final output, both supporting text- and image-to-video with synchronized audio. MiniMax H3 Max and H3 Max Turbo are post-trained variants of MiniMax H3 focused on prompt adherence, aesthetics, and inference performance, with H3 Max also listing reference-to-video.
A few caveats are worth carrying forward. This is a public preview, not general availability — it does not establish access across all tenants and regions. Microsoft says per-model pricing will appear on the Azure pricing page "within the coming weeks," and reporting has flagged that the published unit is inconsistent (a table labeled per video second alongside text referring to per-minute pricing), so confirm the rate in the Foundry portal before planning spend. Data residency varies by model; the initial models are US-hosted, and Microsoft advises organizations with residency requirements to review each model's documented processing and storage terms. Microsoft and fal are scheduled to present the partnership at Microsoft Ignite in session BRK256, "Video AI at Scale with fal on Microsoft Foundry," on November 19.
Foundry solves the hard infrastructure problem — one governed Azure endpoint for several frontier video models, with billing, identity, and compliance handled the way an enterprise expects. What it hands back is a raw MP4. Turning that file into content that actually posts — reframed per platform, captioned in a consistent voice, scheduled, and published — is a separate job, and it's the one [Kompozy](/) is built for.
Take a shot generated from Kling v3 Pro or LTX-2.5 in Foundry and drop it into Kompozy as a source. If you filmed or generated long, the [clipping](/glossary/clipped-short) engine pulls several vertical cuts from it; either way Kompozy reframes to each platform's aspect ratio and burns on-brand, word-synced captions. From that same clip it generates the formats a single model endpoint never will — brand-exact [Carousels](/glossary/hyperframes) and quote cards, per-platform text, a blog article, an email newsletter, and avatar-narrated [Persona Shorts](/glossary/persona-shorts) — every output held to one [Persona Brief](/glossary/persona-brief). Then [Autopilot](/glossary/autopilot) schedules and publishes the set across the eight social platforms plus blog and email, behind a per-post review gate. Foundry is where a developer calls the model; Kompozy is where a creator ships the result — no Azure deployment, no code. For the applied view, see the [Microsoft Foundry fal video breakdown](/ai-tools/microsoft-foundry-fal-video), the honest [review](/reviews/microsoft-foundry-fal-video), and the [platform-vs-product alternative](/alternatives/microsoft-foundry-fal-video).
Five: Kling v3 Pro, LTX-2.5 Fast, LTX-2.5 Pro, MiniMax H3 Max, and MiniMax H3 Max Turbo. They are served by fal and rolled out in public preview in Microsoft Foundry on November 5, 2026. Microsoft says more models are planned.
No. They launched in public preview, not general availability, so access is not guaranteed across all tenants and regions. The initial models are US-hosted, and Microsoft says per-model pricing will appear on the Azure pricing page in the coming weeks — verify rates and regional terms in the Foundry portal before production use.
Most support text-to-video and image-to-video with native or synchronized audio. Kling v3 Pro adds cinematic motion and custom elements; LTX-2.5 ships Fast (speed) and Pro (quality) variants; MiniMax H3 Max and Max Turbo are post-trained for prompt adherence and performance, with H3 Max also listing reference-to-video.
Foundry returns a raw video file; it does not caption, reframe, schedule, or publish. Bring the clip into a content engine like Kompozy, which clips long footage, burns on-brand captions, generates carousels, a blog, and a newsletter from the same source, and auto-publishes across the eight social platforms plus blog and email.