OpenRouter

OpenRouter is a platform that gives developers and teams a single API for hundreds of language, image, video, and speech models from many providers. Requests use an OpenAI-compatible format, and OpenRouter routes each call to a provider, falls back when one fails, and bills the run from one credit balance. A chat interface and free model variants make it usable before writing code.

Key Features

• One OpenAI-compatible endpoint for 500+ models from 80+ providers.
• Automatic routing and fallback across providers when a model fails or is rate limited.
• Unified credit balance and billing across providers, with no markup on token prices.
• Free model variants and a free router with daily request limits.
• Prompt caching, budgets, spend controls, and Management API keys.
• Bring your own keys with a monthly no-fee inference allowance.
• Enterprise options such as SSO/SAML, in-region routing, and zero data retention.

Who is it for

• Developers adding one or many models to an application.
• Teams comparing models or benchmarking prompts.
• Startups that want a single bill across providers.
• Applications that need image, video, and speech models alongside chat.
• Enterprises requiring governance, regional routing, and data policies.

What can you do with OpenRouter

• Send the same request to different models and compare the results.
• Route production traffic with fallbacks so a provider outage does not stop a feature.
• Use free models for prototypes and low-volume tools.
• Track spend by workspace and API key.
• Bring provider keys and route directly when needed.
• Add image, video, speech-to-text, or text-to-speech models through their endpoints.

How does OpenRouter work

1. Sign up and create an API key.
2. Send a chat-completions request in OpenAI format with a model identifier.
3. OpenRouter selects an available provider and applies any routing rules.
4. If a provider fails, fallback routing tries another; failed attempts are not charged for tokens.
5. Track usage in activity logs, and add credits to raise the free-model limits.

Pricing

• Free: 25+ free models, 4 free providers, 50 requests a day, and 5 workspaces.
• Standard: pay-as-you-go at provider list price plus a 5.5% platform fee on credit purchases.
• Business: pay-as-you-go with an 8% platform fee, advanced notifications, and in-region routing.
• Enterprise: contract pricing with SSO/SAML, private endpoints, SLAs, and invoicing.
• BYOK: $25,000 of list-price inference a month without a fee, then 5%.
• The free-model daily cap rises to 1,000 requests after an account buys $10 in credits.

FAQ

Do I need a subscription to use OpenRouter?

No. The Free plan works without a subscription, and paid usage is pay-as-you-go through credits rather than a monthly seat price.

Does OpenRouter mark up model prices?

No. Inference is billed at the provider's list price, and the platform fee is charged when you buy credits, not on individual requests.

What happens if a provider fails?

Fallback routing can try another provider automatically. Under Zero Completion Insurance, a failed attempt is not charged for its model tokens.

Can I use my own provider keys?

Yes. Standard and Business include $25,000 of list-price inference a month through BYOK without an OpenRouter fee, then charge 5%.

Which models are free?

The platform lists more than 25 free models from 4 free providers. Free models are capped at 20 requests per minute and 50 a day by default.

Is there an enterprise plan?

Yes. Enterprise adds SSO/SAML, private endpoints, custom data policies, contractual SLAs, invoicing, and higher rate limits.