---
type: "post"
id: "microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f"
slug: "microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f"
title: "Microsoft Foundry adds Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo via a fal partnership."
description: "Microsoft says five fal-served video models are entering public preview in Microsoft Foundry: Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo."
retrieval_nugget: "Microsoft says five fal-served video models are entering public preview in Microsoft Foundry: Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo. For teams building multimodal products, the change adds model choice inside Foundry, with published per-second rates to test quality, speed, and cost against a specific workflow."
published_at: "2026-11-07"
updated_at: "2026-11-07"
record_date: "2026-11-07"
date_kind: "published_at"
topics: ["models"]
entities: ["Microsoft","Microsoft Foundry","fal","Kling v3 Pro","LTX-2.5 Fast","LTX-2.5 Pro","MiniMax H3 Max","MiniMax H3 Max Turbo"]
related_patterns: ["generative-media-becomes-application-infrastructure"]
novelty_label: "new_fact"
what_changed: "Microsoft Foundry is adding Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo through a new fal partnership."
prior_refs: ["https://newruntime.com/posts/microsoft-launches-decision-1-model-in-foundry-analysis-5322bd/"]
source_url: "https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/building-the-next-generation-of-multimodal-ai-experiences-with-fal-in-microsoft-/4570032"
source_urls: ["https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/building-the-next-generation-of-multimodal-ai-experiences-with-fal-in-microsoft-/4570032"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f"
status: "published"
visuals: []
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f/","markdown":"https://newruntime.com/posts/microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f.md","json":"https://newruntime.com/posts/microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f.json"}
markdown_url: "https://newruntime.com/posts/microsoft-foundry-adds-kling-v3-pro-ltx-2-645f8f.md"
---

# Microsoft Foundry adds Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo via a fal partnership.

## Retrieval answer

Microsoft says five fal-served video models are entering public preview in Microsoft Foundry: Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo. For teams building multimodal products, the change adds model choice inside Foundry, with published per-second rates to test quality, speed, and cost against a specific workflow.

## Summary
Microsoft says five fal-served video models are entering public preview in Microsoft Foundry: Kling v3 Pro, LTX-2.5 Fast and Pro, and MiniMax H3 Max and Max Turbo. For teams building multimodal products, the change adds model choice inside Foundry, with published per-second rates to test quality, speed, and cost against a specific workflow.

## What happened
Microsoft announced on November 5, 2026, that Kling, LTX, and MiniMax video models are rolling out in public preview in Microsoft Foundry through a partnership with fal. Microsoft describes the offerings as a first-party Azure service, served by fal and sourced from the companies that make the models. The initial lineup comprises Kling v3 Pro, LTX-2.5 Fast, LTX-2.5 Pro, MiniMax H3 Max, and MiniMax H3 Max Turbo. Microsoft says it plans to add more models over time.

Microsoft presents the portfolio as text-to-video and image-to-video options for developers and creators. Kling v3 Pro combines cinematic visuals and fluid motion with native audio generation and custom element support. Its image-to-video capability is intended to give developers control over how a starting image becomes a moving scene. LTX-2.5 Fast is the speed-optimized option for rapid iteration and previews; LTX-2.5 Pro is the quality-optimized choice for high-fidelity final output. Both LTX variants support text-to-video, image-to-video, and synchronized audio.

MiniMax H3 Max and MiniMax H3 Max Turbo are post-trained variants of MiniMax H3 focused on prompt adherence, aesthetics, and inference performance, Microsoft says. The listed formats for H3 Max include text-to-video, image-to-video, and reference-to-video, while H3 Max Turbo is listed for text-to-video and image-to-video. Microsoft frames the selection as a way to use different models for different creative requirements rather than one model for every stage: teams exploring concepts may weight speed and cost, while teams producing fewer assets may put more weight on output quality.

The initial portfolio is hosted in the United States, and Microsoft calls Foundry the first provider with a US-hosted Kling 3 endpoint. It says organizations with data-residency requirements should review the documented data-processing and storage terms for each model. Microsoft also points to platform capabilities for evaluation, optimization, governance, security, observability, and infrastructure as teams move toward production.

Microsoft lists rates in US dollars per generated video second. MiniMax H3 Max costs $0.05 at 480p, $0.08 at 768p, and $0.16 at 1080p; H3 Max Turbo costs $0.025, $0.04, and $0.08 at those resolutions. LTX-2.5 Fast costs $0.09 at 720p, $0.13 at 1080p, $0.19 at 1440p, and $0.30 at 4K. LTX-2.5 Pro costs $0.12 at 720p and $0.17 at 1080p. Kling v3 Pro is listed in the US Data Zone at $0.1232 with audio off, $0.1848 with audio on, and $0.2156 with audio and voice control. Microsoft says Azure pricing-page listings will follow in coming weeks.

## What the archive adds
New Runtime's earlier coverage reported that [Kling 4.0 became available through Cloudflare AI Gateway and Workers AI](https://newruntime.com/posts/kling-4-0-video-generation-model-available-on-8479b9/) on October 31, 2026, with per-second pricing and API examples. This is a new distribution event in Microsoft Foundry, rather than the same model release.

## Trend signal
This supports [generative media becomes application infrastructure](https://newruntime.com/patterns/generative-media-becomes-application-infrastructure/): Microsoft is presenting video generation as a set of selectable production components, not as a single standalone creation tool. The announcement makes quality, latency, input modality, hosting, and per-second cost into platform selection variables. It does not, however, establish comparative results or production outcomes for the portfolio.

## Why now
Microsoft says the partnership brings video generation into the broader platform developers use to build AI applications. The company is also directing builders to an Ignite session on November 19, 2026, about video AI at scale with fal on Foundry. The timing places the partnership alongside an effort to make generative-media workloads addressable through Foundry's existing application-building environment.

## Why it matters
For a team adding video features, the practical change is not simply five more model names. It is a larger choice set within a platform the team may already use for AI application development. A workflow might use a faster option for early creative exploration and a quality-oriented option for selected output. Image-to-video and, for some models, reference-to-video also allow an application to begin with an approved product image or other controlled visual input.

The rates create a usable first-pass budget for generation, but teams still need a task-level evaluation. Test representative prompts, inputs, audio requirements, and acceptance criteria; measure latency and the proportion of outputs that are usable; then calculate the cost of accepted clips. The catalog makes this comparison easier to organize, but it does not identify a universal best model.

## Limits and uncertainty
This is a public preview rather than a general-availability commitment. Microsoft's announcement does not provide independent comparative measurements for visual quality, latency, prompt adherence, or reliability across the five models. Hosting is initially in the United States, while processing and storage terms must still be checked model by model against organizational requirements. Published rates are per generated second and vary with resolution and audio configuration; they do not by themselves measure the production cost of retries, rejected generations, storage, or application work around generation.

## Reader takeaway
Microsoft Foundry now offers five fal-served video models in public preview, with concrete per-second prices and distinct speed, quality, audio, and input trade-offs. Treat the release as a structured evaluation opportunity: choose representative jobs, verify data requirements, and compare the cost of accepted output rather than selecting on a headline rate alone.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Models](https://newruntime.com/topics/models.md): Explore the models topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Sources](https://newruntime.com/sources/): See the original sources this archive cites.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)