VEED Dubbing is now live on fal
The Pro tier covers 94 languages and 14 regional accents, keeps background audio, and adds voice cloning strength controls.
Every fal update in one feed — release notes, features and changes, straight from the source.
50 updates
The Pro tier covers 94 languages and 14 regional accents, keeps background audio, and adds voice cloning strength controls.
Kling 4.0 is now publicly available on fal with API access for every customer, bringing 30-second native video with audio and up to 10 keyframes.
The fal plugin lets you generate content with fal’s models directly from a Grok Bot conversation, without switching apps.
Developers can now generate editable SVGs from text, convert images to vectors, and guide results with reference images through fal.
fal now offers Kling 4.0 through its own interface and fal Agent, highlighting steadier complex motion and stronger reference consistency.
The new fal experience runs frontier world models such as LingBot World, letting you build characters, explore environments and direct scenes that react as you go.
The limits cover CLI, Python SDK and REST management calls; requests to app endpoints are unaffected.
Keys can be limited to specific models and set to expire, while existing ADMIN and API keys continue to work.
Creators can use the Playground, Sandbox, and Workflows, while Viewers can inspect request logs and the library without making changes.
Prompts and reference images connect to image, video, audio and 3D models, with access depending on plan and workspace settings.
Pro and Lite video models arrive alongside VSR and VSR Lite super-resolution options.
Gemini 3.8 Flash TTS and Flash Lite TTS bring 30 voices and text-based delivery direction to fal's API.
Recraft's newest image model runs on fal with roughly 1.3-second generation times for faster prompt iteration.
fal's H3 Max model converts Blender previs up to 15 seconds long into photoreal video, guided by optional reference images.
fal's second State of Generative Media report draws on its own platform data to track video editing, faster iteration and a shift toward full environments.
Tripo3D P2 on fal generates textured 3D meshes with four texture-quality settings and separate geometry and texture seeds.
Five built-in styles, including pixel art and VHS, now guide H3 Max video generation from text or reference images.
Meshy V7.1 on fal turns text or multi-view images into textured 3D models with optional PBR maps.
Google DeepMind's Lyria 3.5 joins fal, turning text prompts into music with control over genre, instruments, and vocal delivery.
Grok Voice, the latest speech model from SpaceXAI, is available on fal with 0.70-second responses, word-level transcription timestamps, 30+ voices in 25+ languages, and cloning from two minutes of audio.
fal is running a 50% discount on H3 Max and H3 Max Turbo through September 30, with 768p output at $0.04 and $0.02 per second respectively.
fal Agent gets a $50 per month tier for experimenting, with the full amount returned as fal credits that roll over if unspent.
Creatify's Boreal video model is available on fal, turning prompts into product ads, UGC, and presenter videos with synced speech, music, and sound effects.
Shared WebSocket endpoints on fal can now bill by declared unit counts instead of only by connection duration.
FLUX 3 Video Edit lets fal users restyle shots, swap characters and rewrite dialogue with lip sync from one prompt.
Available to US companies on fal, hosted entirely on US infrastructure, with native audio-video generation and multimodal input control.
H3 Max on fal now generates consistent multi-angle views from a single viewpoint in under three seconds.
Bria's Ad Delayer model reconstructs flat ad images into editable layers with SVG text headlines, now running on fal.
Deployed apps on fal get a dashboard tab for testing any endpoint's input form without leaving the page.
H3 Max Director 1.1 adds mid-generation keyframes, audio conditioning, 1080p output, and continuous generations up to 15 minutes.