Generative media started with developers. Then studios found it, then retail brands, then enterprise teams building things we didn’t initially anticipate. fal now powers generative media workloads for 2.5M+ developers and leading companies including Amazon MGM Studios, Canva, and Adobe, serving millions of daily inference calls with 99.99% uptime. That kind of growth calls for partners who can meet you at scale. Today we announced a strategic partnership with Amazon Web Services (AWS). Read about what this means and where we're headed from our Compute Partnerships lead, Emir Lise. Read the full blog post here: https://lnkd.in/gVDcFQ6X
About us
Generative Media Cloud
- Website
-
https://fal.ai
External link for fal
- Industry
- Technology, Information and Internet
- Company size
- 51-200 employees
- Headquarters
- San Francisco
- Type
- Privately Held
Locations
-
Primary
Get directions
San Francisco, US
Employees at fal
Updates
-
Avatar V is now live on fal! Create studio-quality avatar videos from a 15-second recording with: • Strong realism and identity consistency • Lifelike motion and expressions • Stable across multiple camera angles • Supports long-form performance and dialogue The quality jump in avatar generation over the past year has been wild.
-
A big congratulations to our customer Creatify AI on today's launch. The intersection of AI and creative performance is what makes advertising one of the most exciting spaces right now, and the Creatify team is a great example of the innovation happening in it. Proud to be part of your journey.
Today we’re announcing a major breakthrough in AI video generation: Creatify Agent, the world’s first viral ad orchestrator. Today, every ad maker works the same way. You type in a prompt, one model generates an AI video. The ad looks like AI slop. Creatify Agent makes ads without AI slop because it does not start by generating. It starts by understanding what makes an ad work and the existing content you already have. Drop in a product URL, and Creatify Agent researches the highest-performing ads across TikTok and Instagram in your category, identifies the creative formats already working, and uses them as references for your brand. From there, it writes a direct response script built to sell, then scans your existing content library to find real UGC, product shots, b-roll, and brand assets it can use before generating anything with AI. Once the agent understands the product, the category, and the assets available, it storyboards the ad and uses AI generation to fill the gaps. Even then, Creatify does not rely on one model behind a chat box. It orchestrates across multiple specialized agents and generation models, with different agents handling research, copywriting, storyboarding, asset selection, scene generation, editing, and quality control. That quality-control layer is what makes the system so different. Creatify runs a vision-capable critic inside the production loop, so every ad is checked against your brand before delivery: colors, product details, prices, labels, on-screen text, CTAs, and overall shippability. And if something is wrong, Creatify does not recreate the entire ad. It finds the broken section, fixes that specific piece, drops it back into the editor, and rebuilds the ad locally. That is the difference between an AI ad maker and an AI ad orchestrator. AI ad makers generate. Creatify Agent orchestrates sub agents just like a real human would. Every other company promises to make a “more realistic AI video.” Consumers do not trust fake avatars… they feel like this is AI slop. They trust real people, real products, and creative that feels native to the platforms they are watching on. This is how we built Creatify Agent. My co-founders Yinan (Steven) Na, Chaz Z., and I spent 10 years at Meta and Snapchat building the ad systems the entire industry runs on. We built that knowledge directly into Creatify Agent. And so far, Creatify Agent is working. We’re also releasing VideoAdAgent Bench soon, the first public benchmark for AI ad-video production. On VideoAdAgent Bench, Creatify Agent achieved the lowest hallucination score in the field and outperformed general-purpose video agents and frontier text-to-video models on shippability metrics. Try Creatify Agent free: https://lnkd.in/gNkJ8DES Technical Report: https://lnkd.in/gwezxgwu
-
We're welcoming Sean Bonawitz as our first Head of Trust & Safety at fal. Sean brings years of experience building trust and safety programs across some of the internet's most complex platforms. Learn more about what that work looks like inside generative AI, and why he joined.
Sharing a piece I wrote on how fal approaches trust & safety in generative media. I owe a lot to the people who taught me this craft. The Navy EOD professionals who showed me what operating under real pressure looks like. My colleagues at YouTube, TikTok, Patreon, and TextNow who taught me this discipline by example. The team at fal now, who treat this work as foundational. Grateful to all of you. Genuinely excited about what we're building here. https://lnkd.in/gGj_ZatV
-
Generate images and video directly from your terminal with Claude + our new genmedia CLI. With genmedia CLI you can: • Search and run 1,000+ generative media models from terminal • Use Claude to generate commands and workflows • Run async image, video, audio, and 3D generations • Automatically download structured outputs and local files • Install reusable agent skills for repeatable workflows In this tutorial we walk through: • Installing genmedia CLI • Setting up Claude projects • Generating images with Nano Banana Pro • Running multi-shot video workflows • Using fal Community Skills This is quickly becoming a more powerful way to build with generative media than jumping between browser tabs and model UIs. Watch the full tutorial: https://lnkd.in/gfy-YvGJ Install the genmedia CLI: genmedia.sh
-
Today we’re introducing the genmedia CLI, generative media directly from the command line. Generate images, video, 3D and audio from your terminal, alongside Claude and other AI agents. Built for developers working natively in code, scripts, and automated pipelines. Key features: • Terminal-native media generation • Works alongside AI coding agents • Easy to integrate into scripts + workflows • Installable “skills” for reusable creative workflows and consistent outputs Skills are pre-configured generation setups for things like filmmaking, gaming assets, product visuals, anime, storyboarding, and more, installable in one command. Install here: genmedia.sh Open-source skills repo: https://lnkd.in/gkPTZqet More workflows and skills coming soon 🚀
-
What separates made from generated, and why it matters more as the tools get better, not less. Burkay Gur has been sitting with this since a conversation with Steve Jang at Kindred Ventures.
-
Great conversations have a way of sharpening your thinking. Burkay Gur had the opportunity to sit down with Steve Jang at Upfront Summit to dig into where generative media is headed and it was one of those chats that leaves you more energized than when you walked in. Thank you to Steve and the Kindred Ventures team for the time and the thoughtful exchange. And to Mark Suster, Kerry Bennett, and the Upfront Ventures team for putting us on stage.
Recently at Upfront Summit, Steve Jang sat down with Burkay Gur, co-founder and CEO of fal, to talk about “What’s Coming in Generative Media.” While much of the world focused on chatbots and LLM agents, a parallel world of infrastructure and inference demand in AI has been quietly taking shape. In the areas of design, product development, advertising, marketing, and entertainment, a multitude of state-of-the-art image, audio, and video models are rapidly transforming multiple industries. For the past century, human attention has become increasingly focused on video, images, and audio modalities, beyond textual information. This truth makes the opportunity for AI storytelling and creativity to put control, speed, and fidelity in the hands of everyone. A brave, exciting world. However, on a technical level, generative video simply demands a higher bar of inference quality and far more compute resources than text from LLMs, which could make generative media a dominant driver of NVIDIA GPU workloads alongside agents and robotics. At the same time, the conventional studios, agencies, and consumer brands are speeding up and amplifying creativity and production, enabling high-end output with less resources in an unprecedented compression of traditional timelines. Today, fal serves as the first true Generative Media Cloud, hosting and optimizing serverless inference for the world’s greatest video, image, audio/voice, and world models to serve developers, designers, marketers, studios, and large enterprises, such as Amazon, Adobe, Canva, and Shopify. Some highlights from their talk: ▪️ In the past 2 years since launch: fal has scaled from ~20 models to 1,000+, adding new models weekly ▪️ Today, their platform is used by close to 3M prosumer/consumer developers and servicing Fortune 100 companies ▪️ The areas of design, product development, advertising, and marketing are seeing the earliest adoption and growth ▪️ Entertainment studios are quietly and not-so-quietly using generative media models for internal production workflows ▪️ World models, gaming, and truly multimodal models are rising quickly on the platform ▪️ fal users have generated over 1B pieces of content last month alone using their platform Thanks to Mark Suster, Kerry Bennett, and the Upfront Ventures team for having us onstage and spotlighting the state of the surging generative media wave and @Fal’s fundamental work in the space. https://lnkd.in/gnRBMf2D
What's Coming in Generative Media: Burkay Gur & SteveJang at Upfront Summit 2026
https://www.youtube.com/
-
fal reposted this
Announcing World Model Accelerator, fal's new product line. We've spent years building the inference layer for generative media. In 2023, we shipped 30 FPS, low-latency, action-controlled diffusion APIs, years before anyone else. Fast forward to today and we've been powering foundational model companies across image, video, real-time speech-to-speech and action-controlled world models. Now we're opening it up: World Model Accelerator is the system behind it all. Our in-house inference engine hits SOTA performance on Hopper and Blackwell for Diffusion Transformer workloads, both causal and bi-directional. Built on fal Serverless, it scales seamlessly from 1 GPU to 1,000+. This is the same platform that's been running real workloads, at scale, for years. A new WebRTC gateway optimizes latency between users and GPUs (routing requests to the closest region), built on the same infra that powers our real-time speech workloads. And direct access to the fal Marketplace puts your model in front of enterprises spending hundreds of millions on generative media, with a real co-selling motion behind it. This is the infrastructure layer for world models.
-
-
Announcing World Model Accelerator, fal's new product line. We've spent years building the inference layer for generative media. In 2023, we shipped 30 FPS, low-latency, action-controlled diffusion APIs, years before anyone else. Fast forward to today and we've been powering foundational model companies across image, video, real-time speech-to-speech and action-controlled world models. Now we're opening it up: World Model Accelerator is the system behind it all. Our in-house inference engine hits SOTA performance on Hopper and Blackwell for Diffusion Transformer workloads, both causal and bi-directional. Built on fal Serverless, it scales seamlessly from 1 GPU to 1,000+. This is the same platform that's been running real workloads, at scale, for years. A new WebRTC gateway optimizes latency between users and GPUs (routing requests to the closest region), built on the same infra that powers our real-time speech workloads. And direct access to the fal Marketplace puts your model in front of enterprises spending hundreds of millions on generative media, with a real co-selling motion behind it. This is the infrastructure layer for world models.
-