- Featurefal.ai
Fal.ai introduces platform-level HTTP headers for inference control
Fal.ai added platform-level HTTP headers to control inference request behavior, including timeouts, runner routing, queue priority, media expiration, payload storage, retries, and fallbacks. These hea…
- Featurefal.ai
Fal.ai standardizes common model arguments for image generation APIs
Fal.ai introduced standardized common arguments for its image generation models, including `seed` for reproducibility, `num_images` for batch generation, `image_size`/`aspect_ratio` for dimensions, `e…
- Featurefal.ai
fal.ai introduces configurable media expiration and payload storage controls
fal.ai now allows developers to set custom expiration durations for generated media files via the `X-Fal-Object-Lifecycle-Preference` header, replacing default retention. Request payloads remain store…
- Featurefal.ai
Fal.ai updates concurrency limits for model APIs with automatic scaling
Fal.ai revised its concurrency limits system, where accounts now start with 2 concurrent requests and scale up to 40 automatically based on credit purchases. Limits apply globally but can be increased…
- Featurefal.ai
fal Serverless enables GPU autoscaling for custom AI models
fal Serverless now allows users to deploy custom AI models, pipelines, and applications on GPU infrastructure that autoscales from zero to thousands of machines. Users can migrate existing servers or …
- Featurefal.ai
fal.ai enables publishing serverless apps to Marketplace with shared auth and usage-based billing
fal.ai now allows developers to publish their serverless apps to the Marketplace, enabling any user to call them with their own API key and pay for their own usage. Apps must run in shared auth mode, …
- Launchfal.ai
Fal.ai launches dedicated GPU Compute instances with H100 SXM GPUs
Fal.ai introduced Compute, a new product offering providing dedicated NVIDIA H100 SXM GPU instances for sustained workloads like training and fine-tuning. Two instance types are available: 1xH100 and …
- Featurefal.ai
Fal.ai enhances team management with archiving, roles, and cross-team policies
Fal.ai introduced team archiving, a three-tier role system (Admin, Developer, Billing), and organization-wide policies for team creation, invites, visibility, and SSO. Archiving teams locks accounts a…
- Featurefal.ai
Fal.ai introduces Model Access Controls for enterprise teams
Fal.ai launched Model Access Controls enabling organization admins to restrict which models team members can use via API or Playground. Admins can set models to Allowed, Blocked, or Inherit, with ente…
- Featurefal.ai
Fal.ai expands model consumption and deployment options with updated quickstart guide
Fal.ai now offers two primary paths for AI model integration: consuming 1,000+ production-ready models via Model APIs or deploying custom models using Serverless infrastructure. The quickstart guide p…
- Launchfal.ai
alibaba/happy-horse/image-to-video
Fal.ai introduced Happy Horse 1.0, an AI model converting still images into 1080p videos with synchronized audio, Foley sounds, and multilingual lip-sync. Outputs grant full commercial rights, while p…
- Launchfal.ai
fal-ai/nano-banana
fal.ai introduced nano-banana, a new AI image generation model with customizable input controls. The model charges $0.039 per image, with a $1.00 bundle offering 25 runs. Users can view generation log…
- Launchfal.ai
fal-ai/bytedance/seedream/v4.5/edit
ByteDance’s Seedream v4.5 introduces natural language-driven image editing at $0.04 per edit, supporting up to 10 reference images for complex compositions. The model consolidates generation and editi…
- Launchfal.ai
fal-ai/veo3.1
Google DeepMind’s Veo 3.1, a refined version of Veo 3, now supports true 4K output, native audio generation, and SynthID watermarking. Available via fal.ai’s pay-per-second API, pricing ranges from $0…
- Launchfal.ai
HappyHorse-1.0The Top Ranked AI Video Model
Alibaba’s HappyHorse-1.0, ranked #1 on the Artificial Analysis Video Arena for Text-to-Video and Image-to-Video, is now available via API on fal. The model generates 1080p video with synchronized nati…
- Launchfal.ai
openai/gpt-image-2/edit
Fal.ai launched an image editing API for OpenAI's GPT Image 2 model, enabling fine-grained edits to existing images via natural language prompts or optional masks. The endpoint supports real-time stre…
- Launchfal.ai
ChatGPT Images 2.0Quality-First Image Generation
OpenAI launched ChatGPT Images 2.0 on April 21, 2026, a next-generation image model prioritizing photorealism, pixel-perfect text rendering, and brand-consistent product photography. The model introdu…
- Launchblog.fal.ai
Introducing PATINA
AI-generated images often lack the structured, usable material properties required for traditional CGI workflows, such as PBR (Physically Based Rendering) maps like basecolor, roughness, or normal map…
- fal.ai
bytedance/seedance-2.0/fast/text-to-video
ByteDance has introduced Seedance 2.0 Fast, a text-to-video generation model available via Fal AI, with a new pricing model. The service charges $0.2419 per second for 720p video generation, calculate…
- Launchfal.ai
Seedance 2.0APIUnlock The Next Generation of AI Video
ByteDance has launched Seedance 2.0, its most advanced video generation model, now available via API through fal.ai as of April 10, 2026. The model introduces cinematic video generation with native au…
- fal.ai
Sandbox | fal.ai
The provided content is a placeholder or empty state for a product page labeled 'Fal.ai Sandbox,' showing no active content or generated output. The interface suggests users can configure settings and…
- Launchfal.ai
Happy OysterThe Open-Ended AI World Model
fal introduced Happy Oyster, an open-ended AI world model enabling real-time 3D world creation and interaction via text, voice, or images. Unlike traditional video generators, it builds and responds t…
Track Fal on autopilot
- · Weekly AI brief: narrative summary of what shipped, every Monday 9 AM
- · Email or Slack alerts, or chat with the archive in your dashboard
- · Add Fal + up to 2 more competitors free, no credit card