Features
Zero-shot object detection from plain text
Custom model creation in under 5 minutes
Instant deployment via managed API without infrastructure
Attribute Detection for extracting structured object details (beta)
Conversational Model Builder for model definition (beta)
Video processing at up to full frame rate
Image classification and object detection
Playground for interactive model testing
Model Templates for quick customization
Bespoke model finetuning for volume customers (beta)
Video billed at 5 FPS base rate
Sketch-to-image diffusion generation
Optional text prompt for style guidance
Client-side processing via Replicate API
Multiple output variations per sketch
No account or sign-up required
Open-source codebase on GitHub
Real-time preview of generated results
Unlimited free generations
512x512 output resolution
Privacy-focused – no server-side logs
Self-hosting option via GitHub
Lightweight web interface
