OpenRouter
FreeUnified API + live pricing across ~300 models.
FreeFree tier
Inputs: textOutputs: text
About OpenRouter
OpenRouter is a unified API and marketplace that provides access to over 400 AI models from 70+ providers. Launched in early 2023 as the first LLM marketplace, it eliminates vendor lock-in while offering live pricing, auto-routing, budgets, spend controls, prompt caching, and enterprise-grade reliability. Developers can compare models by price, context, and benchmarks, and choose between free, pay-as-you-go, or enterprise plans. OpenRouter processes over 200 trillion monthly tokens and serves 10 million global users, backed by leading investors including Alphabet.
Key Features
Access to 400+ AI models from 70+ providers
Live pricing comparison across models
Auto-routing and preferred vendor selection
Budgets, spend controls, and prompt caching
Management API keys and admin controls
Enterprise features: SSO/SAML, contractual SLAs, BYOK, data policy-based routing
Free tier with 50 requests/day and 25+ free models
Pay-as-you-go with no minimum spend (5.5% fee) and volume discounts on enterprise
Pros & Cons
Pros
- Massive model selection from 70+ providers in one API
- Live pricing allows cost-effective model selection
- No vendor lock-in; easy to switch or combine models
- Free tier available for experimentation
- Enterprise-grade reliability and uptime monitoring
- Auto-routing improves resilience and latency
Cons
- Free tier limited to 50 requests/day and free models only
- Pay-as-you-go includes a 5.5% platform fee on top of provider pricing
- Limited documentation on custom routing policies for free tier
- Dependency on third-party providers for model availability
Best For
Unified access to multiple LLMs for AI-native startups and indie hackersEnterprise AI gateway with cost optimization and vendor flexibilityComparing model performance and pricing for specific tasksBuilding multi-provider AI applications without individual API integrationsManaging spend and usage across teams with budgets and activity logs
FAQ
How are tokens billed?
Tokens are billed based on the model's provider pricing plus applicable platform fees. Pay-as-you-go includes a 5.5% fee on list price; enterprise plans have volume discounts.
Do you mark up provider pricing?
OpenRouter applies a platform fee: 5.5% on pay-as-you-go, with discounts available for higher volume. Free tier only includes free models.
Are failed or fallback attempts billed?
Failed or fallback attempts are not billed. Only successful responses consume tokens.
Do you enforce rate limits?
Yes. The free tier has 50 requests per day. Pay-as-you-go and enterprise have higher or optional dedicated limits.
Do you support SSO?
Yes, enterprise plans support SSO/SAML.
Do you train on customer data?
No, OpenRouter does not train on customer data. Privacy and security policies are in place.