In any modern AI creative workflow, upscalers play a critical post-processing role. They take the raw generations from diffusion or generative video models (often limited to 720p resolution) and convert them into production-ready assets suitable for film, design, or print.

There have been a lot of AI conferences, but none focused on generative media. Until now. On October 24, 2025, inside the Ferry Building overlooking the San Francisco bay, we hosted the inaugural Generative Media Conference. 300 of the leading founders, researchers, studio heads, architects, advertisers, and investors in the

Today, we’re excited to announce that fal has acquired Remade, a YC-backed startup building powerful tools for generative media creation. Remade has spent the last two years pushing the boundaries of what’s possible with generative media, training custom models, building unique product experiences and contributing to the open-source

We’re thrilled to announce that Sora 2 and GPT Image 1 are now available on fal.ai. These groundbreaking generative AI models push the boundaries of what’s possible in video and image creation. Whether you’re building creative tools, social media content, or interactive experiences, these new releases

What Changed Google released Veo 3.1 in mid-October 2025. Both versions run on identical veo-3.0-generate-001 architecture. This is refinement, not reinvention. Improvements come from better training data and enhanced post-processing, which shapes everything about the upgrade decision. Three material improvements define this release: Synchronized audio generation directly from

We’re excited to introduce the Sima Labs image and video upscaler, now available on fal. This new upscaling suite brings crystal-clear enhancement to both images and video, powered by advanced AI models that preserve fidelity, colour, and detail at every level.

Strategic Leap Forward Tencent's Hunyuan Image 3.0 represents a significant advancement in open-source text-to-image generation. With 80 billion parameters (13 billion active through MoE architecture), it's now the largest available open-source model in this space. But size alone isn't the story. The core

We're excited to launch Bria's latest text-to-image model, FIBO, on fal! FIBO turns simple text prompts into structured JSON schemas, giving creators predictable, controllable visuals at scale. Trained on richly detailed captions (often 1,000+ words), it deeply understands lighting, composition, color, camera settings, and depth

We’re thrilled to announce that MiniMax Hailuo 2.3 is now available on fal on day 0. This release delivers a major leap in video realism, camera control, and motion physics, making it MinMax's most capable generative video model to date. The model is available through Pro,

When using machine learning models in the real world, performance isn’t just about how fast your GPU can crunch numbers — it’s also about how quickly you can get your model there. Every second spent waiting on a checkpoint to load is a second your GPUs sit idle instead

We’re thrilled to announce that LTX 2.0 is now live on fal at day 0. LTX 2.0 brings next-level text-to-video and image-to-video generation, offering a blend of speed, fidelity, and flexibility for creators of all kinds. With Fast and Pro options supporting 1080p, 1440p, and 4K output,

Kling 2.5 Turbo Standard brings faster, more stable motion and tighter prompt adherence to image-to-video. It’s ideal for cinematic shots, product visuals, sports, and narrative scenes. The model builds on the 2.5 generation with upgrades in temporal consistency, camera control, and style stability, while reducing generation costs

We’re excited to introduce maestro, Beatoven.ai’s proprietary model for high-quality music and sound-effect generation - now available on fal.  Designed for creators, developers, and studios, maestro brings fast, high-fidelity audio generation directly to your workflow through two powerful APIs: Music and Sound Effects generation. Music Generation API

We are excited to announce the release of Reve now available from day 0 on fal. Reve is a state-of-the-art image generation model that brings visual intelligence, layout design, and editing into one seamless creative workflow. With text-to-image, image editing, and multi-reference generation, Reve gives creators precise control over composition,

We’re thrilled to introduce Meshy, now available on fal. Meshy is the newest addition to fal’s growing suite of generative media models. It's a powerful 3D generation model that turns words or images into high-quality 3D meshes ready for rendering, animation, and real-time applications. With text-to-3D,

We are pleased to announce the release of Veo 3.1, now available on fal from day 0. This latest evolution of Google DeepMind’s Veo model series brings a new level of cinematic control to generation, combining expressive visuals with native audio generation for synchronized, sound-integrated storytelling straight out

Moondream 3 Preview is now live on fal! The model is designed for real-world vision tasks from drones and robotics to medical imaging and retail. In this update, Moondream got a larger context window and has become much smarter. The model can now understand more sophisticated prompts and produce longer,

We’re excited to introduce Ovi, now available on fal on day 0. Built by Character AI, Ovi is the first open-source video model with native audio generation, delivering synchronized picture and sound from a single prompt with no post-production required. To help you get started, use coupon code Ovi

Develop and scale real-time generative image, video, audio, and vision apps faster with fal on Google Cloud Marketplace. Today, we’re excited to share that fal’s Generative Media Platform is now available on Google Cloud Marketplace. Teams can evaluate and purchase our production-ready model APIs directly through Google Cloud,

We’re excited to introduce Hunyuan Image 3, the next leap forward in open-source text-to-image generation. Built for creators with a vision, this model transforms words into visuals with unprecedented accuracy and artistry. Whether you’re aiming for photorealistic renders, experimental aesthetics, or characters captured mid-motion, Hunyuan Image 3 delivers

Image-to-video tools are everywhere right now. VEED's latest model, Fabric 1.0 allows you to generate faster and longer video outputs. It can animate any image, giving end-users the flexibility to go beyond preset avatars and create a wide variety of content. What is VEED Fabric 1.0?

We are excited to announce the release of Wan 2.5 now available from day 0 on fal. This latest model in the Wan family introduces new capabilities for both text-to-video and image-to-video generation. For the first time ever, Wan videos come alive with native audio generation, opening the door

We are pleased to announce the release of Kling 2.5 Turbo Pro, now available exclusively on fal. This latest model in the Kling family introduces enhanced capabilities for both text-to-video and image-to-video generation, expanding creative possibilities across a wide range of use cases. Kling 2.5 is designed for

While AI has transformed digital interactions, the physical world - with its dynamic environments and real-time demands - has remained out of reach. That’s why we’re excited to announce our partnership with Perceptron. Founded by the team behind Meta’s Chameleon multimodal models, Perceptron introduces breakthrough capabilities in