@modelrelay

Like OpenRouter, but with customer billing and subscriptions for your apps.

modelrelay.ai
Joined August 2025
  • Tweets27
  • Following2
  • Followers60
3Media
ModelRelay is like OpenRouter + Stripe with customer billing in one platform. You no longer need a backend for your AI. ModelRelay does it all for you. Sign up now for 0% platform fee.
1
142
Added Clairic by @Cloudflare, the first extraction model on ModelRelay. It's great at tasks like pulling totals and dates from invoices and receipts, and is super fast like Jev.
54
Added Mercury 3 by @_inception_ai, the first diffusion model on ModelRelay. Get results instantly (< 0.5s). No more waiting for output tokens. Currently 50% off.
20
Added DeepSeek V4.1 Flash 1017. With an Artificial Analysis score of 45.1, it has a great balance of cost, intelligence, and speed. Included with free credits on new accounts.
1
282
ModelRelay now has a ZDR (zero data retention) filter so you can choose model providers that respect your privacy.
1
312
Gemini 4 Aragon and Muse Wonder are now available in ModelRelay. ModelRelay is a good way to try them out before buying a subscription to Meta or Google AI.
87
ModelRelay now includes fast speech-to-text and text-to-speech models Grok Voice Transcribe 2.0 and Gemini 3.8 Flash TTS. Go build some real-time audio apps!
90
Qwen3.8 27B is live on ModelRelay at a blazing 1,500 tokens per second, 10x what you get on a normal GPU host. Cheapest output we found anywhere: $1.49 per million.
61
Built on @modelrelay
Shane Vitarana 📐
@shan3vAug 20
Today I’m launching Cozy, a macOS app that lets you chat with your iMessage history and draft replies that sound like you. I think it does it better than anything else out there currently.
72
Link: modelrelay.ai

ModelRelay | One API for AI models, metered per customer

Call leading AI models through one API, meter every request to the customer who made it, and charge them for it.

modelrelay.ai
7
Grok 4.6 comes out on top as the best model balancing cost and intelligence. Also it's fun playing with the speed, cost, and intelligence sliders on @modelrelay.
1
7
443
ModelRelay now supports the Anthropic Messages API at /v1/messages. Use your existing Anthropic SDK or tooling with ModelRelay—both streaming and unary responses work out of the box. docs.modelrelay.ai/integrati…

SDK Compatibility

Use OpenAI and Anthropic SDKs with ModelRelay

docs.modelrelay.ai
2
330
ModelRelay now supports the OpenAI Responses API. Use your existing OpenAI SDK with ModelRelay as the backend—just point to /v1/responses and authenticate with your ModelRelay key.
1
286
ModelRelay now supports promotional credits for pay-as-you-go tiers. Credits apply automatically when customers subscribe through managed billing, giving you the same incentive options across all billing models.
1
182
ModelRelay now supports promotional credits on tiers. Apply one-time promo credits that integrate seamlessly with your existing billing flow. This is a great way for your customers to try before they buy.
138
New workflow node: sql.validate. Add SQL validation directly to your workflows with policy and profile support. Very useful for building SQL tool loops.
99
New accounts now get free credits and a ready-to-use API key on signup. Start testing immediately. No setup or payment method required.
60
ModelRelay now supports Google OAuth for sign-in. You can authenticate with your Google account or Google Workspace identity starting today.
58
ModelRelay now supports synchronous batch completions. Process multiple LLM requests in a single call with configurable concurrency, fail-fast behavior, and consolidated usage reporting. This is very useful for RLM requests. Updated SDKs available: github.com/modelrelay/sdks
1
59
Load more