# Weyl

> Weyl is purpose-built inference infrastructure for generative media. We provide sub-100ms latency for diffusion models running on Blackwell architecture with FP4 precision.

Key capabilities:
- **Low Latency**: Sub-100ms p99 latency with optimized CUDA kernels
- **Cost Optimized**: FP4 quantization delivers 4x throughput improvement
- **Dual Tiers**: Sync for real-time, Async for cost optimization
- **Advanced Models**: FLUX.2, FLUX.1, Z-Image Turbo, WAN Video

## Getting Started

- [Introduction](https://weyl.ai/getting-started/): Get started with Weyl inference infrastructure
- [Quick Start](https://weyl.ai/getting-started/quick-start/): Get up and running in 5 minutes
- [Authentication](https://weyl.ai/getting-started/auth/): Set up your API keys

## AI Workflows

- [AI Workflows Overview](https://weyl.ai/workflows/): Generate images in Cursor, Claude, v0, Lovable, and Bolt
- [Cursor IDE](https://weyl.ai/workflows/cursor/): AI image generation in Cursor IDE for vibe coding
- [Claude Projects](https://weyl.ai/workflows/claude/): Claude Projects and MCP integration for image generation
- [v0.dev](https://weyl.ai/workflows/v0/): Add AI image generation to v0.dev components
- [Lovable](https://weyl.ai/workflows/lovable/): Full-stack apps with AI image generation on Lovable.dev
- [Bolt.new](https://weyl.ai/workflows/bolt/): Rapid prototyping with AI images in Bolt.new

## API Overview

- [API Overview](https://weyl.ai/api/): Generative media at the speed of thought
- [Core Concepts](https://weyl.ai/api/concepts/): Understanding model families, backends, and formats
- [API Authentication](https://weyl.ai/api/authentication/): Authentication methods and best practices

## Sync Tier

- [Sync Overview](https://weyl.ai/api/sync/): Real-time generation with dedicated capacity
- [Video Generation](https://weyl.ai/api/sync/video/): Sync video generation endpoints
- [Image Generation](https://weyl.ai/api/sync/image/): Sync image generation endpoints
- [Capacity Management](https://weyl.ai/api/sync/capacity/): Check capacity and handle 503 responses

## Async Tier

- [Async Overview](https://weyl.ai/api/async/): Queue-backed generation for cost optimization
- [Queue Submission](https://weyl.ai/api/async/queue/): Submit jobs to the async queue
- [Job Management](https://weyl.ai/api/async/jobs/): Poll, cancel, and manage async jobs
- [Server-Sent Events](https://weyl.ai/api/async/sse/): Real-time job updates via SSE

## Models

- [Models Overview](https://weyl.ai/api/models/): Available models and capabilities
- [FLUX Models](https://weyl.ai/api/models/flux/): Black Forest Labs FLUX.1 and FLUX.2
- [Z-Image Turbo](https://weyl.ai/api/models/zimage/): Alibaba Tongyi Z-Image models
- [WAN Video](https://weyl.ai/api/models/wan/): WAN 2.2 video generation
- [Format Reference](https://weyl.ai/api/models/formats/): Available dimensions and aspect ratios
- [Backend Comparison](https://weyl.ai/api/models/backends/): nunchaku, torch, and tensorrt backends

## Advanced

- [Samplers](https://weyl.ai/api/advanced/samplers/): Sampling methods and configuration
- [Schedulers](https://weyl.ai/api/advanced/schedulers/): Noise scheduling strategies
- [Guidance Tuning](https://weyl.ai/api/advanced/guidance/): CFG and guidance scale optimization
- [LoRA Adapters](https://weyl.ai/api/advanced/loras/): Using LoRA for style and concept injection
- [Detail Enhancement](https://weyl.ai/api/advanced/detail/): Techniques for improving output quality

## WebSocket

- [WebSocket Overview](https://weyl.ai/api/websocket/): Real-time bidirectional communication
- [Sync WebSocket](https://weyl.ai/api/websocket/sync/): Streaming sync tier generation
- [Async WebSocket](https://weyl.ai/api/websocket/async/): Job updates via WebSocket
- [Protocol Reference](https://weyl.ai/api/websocket/protocol/): WebSocket message format and events

## Reference

- [Request Schemas](https://weyl.ai/api/reference/requests/): Complete request body schemas
- [Response Schemas](https://weyl.ai/api/reference/responses/): Response formats and CDN headers
- [Type Reference](https://weyl.ai/api/reference/types/): TypeScript type definitions
- [Error Reference](https://weyl.ai/api/reference/errors/): Error codes and troubleshooting

## Infrastructure

- [Image Uploads](https://weyl.ai/api/infrastructure/uploads/): Upload large images for stable URLs
- [Model Discovery](https://weyl.ai/api/infrastructure/discovery/): List available models dynamically
- [Model Aliases](https://weyl.ai/api/infrastructure/aliases/): HuggingFace model ID resolution

## Optional

- [Backend Comparison](https://weyl.ai/api/models/backends/): Deep dive into nunchaku, torch, and tensorrt
- [Error Reference](https://weyl.ai/api/reference/errors/): Complete error codes and troubleshooting guide
- [Detail Enhancement](https://weyl.ai/api/advanced/detail/): Advanced techniques for quality optimization