# RelayGPU

> RelayGPU is a unified AI compute API that lets developers run Chat, Image, Speech, and Multimodal AI models with a single API call, routing workloads across direct and decentralized GPU infrastructure. Same frontier models at 15-55% below direct API pricing.

## About

RelayGPU provides a single endpoint for accessing 40+ frontier AI models including GPT-5, Claude 4, DeepSeek V3, Gemini, FLUX, Kling, Wan, Sora, and Whisper. It supports two compute paths: Direct Mode for low-latency trusted hosting and OpenGPU Mode for decentralized GPU resources where providers compete in real-time pricing. Developers can switch modes or let Relay choose automatically.

RelayGPU is operated by OpenGPU, Inc. (d/b/a Relay), a Delaware corporation.

## Key Features

- Unified API for Chat, Image, Speech, and Multimodal AI models
- 100% OpenAI SDK compatible - change only the base_url
- Dual engine infrastructure: Direct Mode and OpenGPU Mode
- 15-55% cheaper than direct API providers
- Pay-as-you-go billing with $10 free credits for new users
- Support for JavaScript, Python, Go, and cURL
- API key management dashboard with usage analytics
- Web3 wallet, Google OAuth, and email authentication
- EU Sovereign tier with full GDPR and EU AI Act compliance
- VS Code extension for Claude Code integration

## Pricing

- DeepSeek V3: from $0.25 per 1M tokens
- GPT-5: 40% below OpenAI list price
- Claude 4: 15% below Anthropic list price
- OpenGPU Mode: from $0.02 per 1M tokens for open-weight models
- No minimum commitments, pay only for what you use

## Integrations

- VS Code Extension: One-click Claude Code routing with live savings counter
- AI Agent Frameworks: Compatible with OpenClaw, LangChain, CrewAI, AutoGen, LlamaIndex, Vercel AI SDK, n8n, Flowise
- EU Sovereign: True EU jurisdiction, CLOUD Act proof, zero data retention, ISO 27001, SOC 2

## Documentation

- API Documentation: https://docs.relaygpu.com
- Main Website: https://relaygpu.com

## Pages

### Public Pages
- Home: https://relaygpu.com
- Pricing: https://relaygpu.com/pricing
- FAQ: https://relaygpu.com/faq
- OpenGPU Network: https://relaygpu.com/opengpu
- VS Code Integration: https://relaygpu.com/integrations/vscode
- EU Sovereign: https://relaygpu.com/integrations/sovereign
- AI Agents Integration: https://relaygpu.com/integrations/agents/openclaw
- Privacy Policy: https://relaygpu.com/legals/privacy
- Terms of Service: https://relaygpu.com/legals/terms

### Dashboard (authenticated)
- Playground: https://relaygpu.com/dashboard
- API Keys: https://relaygpu.com/dashboard/keys
- Models: https://relaygpu.com/dashboard/models
- Usage Analytics: https://relaygpu.com/dashboard/usage
- Credits: https://relaygpu.com/dashboard/credits
- Profile: https://relaygpu.com/dashboard/profile

## Products

- RelayGPU API: Unified AI inference API supporting 40+ models across Chat, Image, Speech, and Multimodal types
- OpenGPU Network: Decentralized GPU compute mesh with community contributors and real-time pricing
- Dashboard & Playground: Interactive interface for testing AI models, managing API keys, and monitoring usage
- VS Code Extension: Seamless Claude Code integration with automatic cost savings

## Getting Started

1. Sign up at https://relaygpu.com/dashboard using email, Google, or Web3 wallet
2. Generate an API key from the Keys section
3. Set your base_url to https://relaygpu.com/v2
4. Start making API calls with $10 in free credits

## Community

- Documentation: https://docs.relaygpu.com
- X (Twitter): https://x.com/relaygpu
- GitHub: https://github.com/relaygpu