fal

fal launched FLUX 3 Max, a video model post-trained with Black Forest Labs, making 10-second clips in about 12 seconds, at 75% off for three weeks.

@fal on X

fal launched fal Worlds, a way to create and interact with AI-generated worlds in real time, including models like LingBot World and H3 Max Director.

@fal on X

fal's Flux 3 Video and Upscaler now support 2K and 4K video across text, image and video workflows, with 5-second 4K clips in about 2 minutes.

@fal on X

fal launched fal Reframe, which changes a video's aspect ratio, such as 9:16 to widescreen, by extending the scene instead of cropping.

@fal on X

fal added Bria's Guided Remove Background, which uses a text prompt to keep part of an image and removes the rest with transparency.

@fal on X

fal added Vidu Q4, which makes image-to-video and reference-to-video clips of 3 to 16 seconds with native audio, up to 4K.

@fal on X

fal launched H3 Max Relight, which changes the lighting of an uploaded video using a built-in lighting studio while keeping motion and audio.

@fal on X

fal added Kandinsky 6.0 in Pro and Lite versions, with synchronized audio, built-in upscaling and standalone video super-resolution.

@fal on X

fal is now available in ChatGPT and Codex, where users can generate media, browse it in chat and access their fal Media Library.

@fal on X

fal released a Claude Connector and a GPT plugin so users can generate images and videos inside Claude or ChatGPT Codex without an API key.

@fal on X

fal added Google's Nano Banana 2.1 image model, with faster generation, better subject consistency and mask-based editing.

@fal on X

fal's H3 Max Reference-to-Video now supports first, middle and last frame control, plus image or video references to guide the scene.

@fal on X

fal added FLUX 3 Image from Black Forest Labs, which edits precisely, outputs native 2K and 4K, and combines up to 10 reference images.

@fal on X

fal launched Recast, which swaps the character in an existing video using a reference photo while keeping the original motion, cuts and audio.

@fal on X

fal added Grok Imagine Video 1.5 Lite, a lightweight text-to-video and image-to-video model with native audio and up to 1080p output.

@fal on X

fal added VEED Clean Audio, which removes background noise from speech in audio or video files up to 30 minutes long.

@fal on X

fal added Pixelcut Looping Video, which turns one product photo into a seamless 5 to 15 second loop up to 1080p.

@fal on X

fal added Ideogram 4.5, an image model for text-to-image and editing with accurate typography, mask editing and up to 4 reference images.

@fal on X

fal added Video Insert to H3 Max, which generates a new scene from a prompt or references and inserts it into existing video.

@fal on X

fal Agent is now available to everyone, with no subscription required, for creating image, video, audio and 3D work in one interface.

@fal on X

fal now supports 1080p and 4K video generation on US-hosted Seedance 2.0 for customers using US-hosted infrastructure.

@fal on X

fal added ElevenLabs' Eleven v4 and Eleven v4 Turbo text-to-speech models, with inline audio tags, 100 languages and a low-latency option.

@fal on X

fal added Gemini 3.8 Flash TTS and Gemini 3.8 Flash Lite TTS, with custom voice design and multi-speaker speech in 100+ languages.

@fal on X

fal added native 1080p generation to Seedance US 2.5, with finer detail and cleaner textures.

@fal on X

fal launched H3 Max Extend Video, which extends an uploaded video of up to 15 seconds to 30 seconds.

@fal on X

fal added Recraft V4.1 Flash, an image model from Recraft that generates images in about 1.3 seconds.

@fal on X

fal added 3D to Video on H3 Max, turning Blender previs of up to 15 seconds into photoreal video with reference images.

@fal on X

fal added Meshy V7.1, which creates 3D assets with PBR materials from text, an image or multiple reference views.

@fal on X

fal added H3 Max Styles, letting users generate video in pixel art, hand-drawn, low-poly, 70s retro toon or VHS looks.

@fal on X

fal added Tripo3D P2, which generates textured 3D meshes from text or images with four texture-quality settings.

@fal on X

fal made Seedance 2.5 available on US-hosted infrastructure for US customers, covering text, image and reference to video.

@fal on X

fal added Lyria 3.5 from Google DeepMind, which turns detailed text prompts into music with vocals and mixed genres.

@fal on X

fal added Bria's Product Holding and Virtual Try-On models for product-in-hand imagery and multi-garment changes.

@fal on X

fal launched H3 Max Lip Sync, which turns one photo and one audio clip into a lip-synced video in a median 11 seconds.

@fal on X

fal added Grok Voice, a speech model from SpaceX AI that offers transcription, text to speech in 30+ voices, and voice cloning from two minutes of audio.

@fal on X

fal added Boreal by Creatify, which turns a prompt into product ads, UGC and presenter videos with synchronized audio, up to 20 seconds at 2K.

@fal on X

fal added a $50 per month tier to fal Agent for people who want to experiment, with the full amount returned as fal credits.

@fal on X

fal launched H3 Max Camera Controls, which turns a single viewpoint into a navigable 3D scene in under 3 seconds using set camera angles.

@fal on X

fal made Dreamina Seedance 2.0 available to US companies, with inference running entirely on US infrastructure; access is through sales@fal.ai.

@fal on X

fal added FLUX 3 Video Edit, which changes characters, backgrounds, style and dialogue in a clip from one prompt, with lip sync following new dialogue.

@fal on X

fal made H3 Max Reference-to-Video available, letting users generate video guided by reference inputs.

@fal on X

fal added Ad Delayer from Bria, which turns a finished flat ad into a layered, editable composition with headlines as SVG text.

@fal on X

fal added multi-angle generation to H3 Max, letting users set horizontal and vertical camera angles to render a single image from other viewpoints in under 3 seconds.

@fal on X

fal added Phota Identity v2 from PhotaLabs, which builds a persistent personal identity from as few as 5 photos for use across supported image models.

@fal on X

fal added GPT Image 2.5, an image model with precise edits and consistent reference subjects across long editing sessions.

@fal on X

fal released H3 Max Director 1.1, adding keyframes, audio conditioning at any point, native 1080p, and continuous generations up to 15 minutes.

@fal on X

fal's H3 Max and H3 Max Turbo video models now generate native Full HD (1080p) in real time, with 5 seconds of video in under 5 seconds.

@fal on X

fal made reference-to-video on H3 Max generally available, up to 2x faster, with real-time generation now working with up to 4 references.

@fal on X

fal released H3 Max Director through its API, a video model that generates one continuous stream for long-form, action-controlled video, with 75% off for two weeks.

@fal on X

fal launched H3 Max Turbo in preview, a video model running at twice the speed and half the cost of H3 Max, at a promotional price.

@fal on X

fal added Bria's Fibo Edit 1.5, which accepts up to four reference images for product placement, virtual try-on and background replacement.

@fal on X

fal added reference-to-video for MiniMax H3 Max as an early preview, running at roughly real-time speed at 768p.

@fal on X

fal added Gemini Omni Flash 1.1, which handles text, image and video together and supports text-to-video, image-to-video, reference-to-video and editing.

@fal on X

fal added Recraft V4 Styles, which holds a style reference across generations and outputs raster or vector images, in Precise or Flexible mode.

@fal on X

fal launched H3 Max, its own post-trained video model that makes a 5-second 720p clip in under 3 seconds, at half price for a week.

@fal on X

fal added Meta's Muse Image, an image model that plans compositions and self-corrects, with text to image and image edit endpoints.

@fal on X

fal added Wan 3.0, a video model that makes native 30-second clips and accepts text, image, audio, video and document references.

@fal on X