
OpenRouter
6.5/10OpenRouter is a unified API gateway that lets developers call 600+ large language models from 80+ providers with one key, with routing and fallback.
Data last updated November 2026 · Our methodology
Overview
OpenRouter is an inference gateway. It does not rent out servers. It puts one OpenAI-compatible endpoint and one API key in front of more than 600 models from more than 80 providers. Requests can be auto-routed by cost or performance, and if a provider degrades they fall back to another model or provider. OpenRouter also publishes public model rankings with real-world price, latency and throughput data. It suits developers, startups and SaaS teams that want to compare or switch models without writing a separate integration for each lab. It is especially useful while a team is still choosing models. There are no seat or subscription prices. Inference is billed at each provider's list price. Standard and Business accounts buy prepaid credits and pay a platform fee on each purchase: 5.5 percent on Standard and 8 percent on Business. Business adds up to 1,000 workspaces, EU and US in-region routing, and Workload Identity Federation. Enterprise has discounted fees, invoicing, SSO/SAML, contractual SLAs and optional private endpoints. A free tier offers 25+ free models with tight daily request caps. Bring-your-own-key use is free up to 25,000 dollars a month of list-price inference, then costs 5 percent. The main limitation is that it is a hosted SaaS layer sitting between you and the providers, with no private deployment outside Enterprise add-ons. In February 2026 it had two outages of about 35 to 38 minutes each, caused by a caching-layer failure. Users complain about slow support, often through Discord, and about the extra cost of the fee at production volume.
Our Rating
AI Software6.9/10
Hosting & Infrastructure6.2/10
Who OpenRouter is best for
Best software for DevelopersGood fit
One OpenAI-compatible API covers hundreds of models, which makes it built for developers who experiment with LLMs.
Pay-as-you-go credits with no minimum suit early teams that are still choosing models.
Fallback routing and workspaces help SaaS products that embed AI, though the fee adds up at scale.
Less suited
SSO, SLAs and invoicing are available, but only on a contract tier, and reviewers note limited compliance detail.
Pros & Cons
Pros
- A single OpenAI-compatible endpoint and key reach 600+ models, so teams can switch models without new integrations.
- Inference is billed at providers' list prices, and failed or fallback attempts are not charged for tokens.
- Automatic fallback reroutes requests when a provider fails or degrades.
- No minimums or lock-in on the Standard and Business pay-as-you-go plans.
- Published detailed post-mortems after its 2026 outages and added circuit breakers in response.
Cons
- The 5.5% or 8% fee on credit purchases adds cost at scale. One estimate puts it at about 550 dollars a month on 10,000 dollars of inference.
- Support is slow according to reviewers, and Discord is often the main channel. Trustpilot showed 1.7/5 from 41 reviews in May 2026.
- API outages of 38 and 35 minutes on February 17 and 19, 2026 returned errors across all endpoints.
- GitHub issues report that some provider rate-limit errors reach users instead of triggering failover.
- SSO/SAML and contractual SLAs are limited to the Enterprise contract tier.
Key Features
Not included
- Managed WordPress
- Dedicated Servers
- VPS Hosting
- Shared Hosting
OpenRouter Pricing
Free
- 5 workspaces
- Community support
- 25+ free models from 4 providers
- 50 requests per day on free models (1,000 per day after a 10 dollar credit purchase)
Business
- Email support
- 1,000 workspaces
- Advanced notifications
- EU and US in-region routing
- Workload Identity Federation
- Pay as you go with an 8% fee on credit purchases
Enterprise
- SSO/SAML
- Invoicing
- Contractual SLAs
- Higher rate limits
- Dedicated support team
- Managed policy enforcement
- Discounted platform fee by contract
Standard
- 5 workspaces
- Email support
- 600+ models, 80+ providers
- Management API keys and activity log export
- Pay as you go with a 5.5% fee on credit purchases
- Budgets, spend controls, auto-routing, prompt caching
Frequently Asked Questions
No. You pay provider list prices for tokens plus a platform fee when you buy credits: 5.5% on Standard and 8% on Business. Enterprise fees are negotiated.
Yes. BYOK is free up to 25,000 dollars a month of list-price inference, then a 5% fee applies.
Yes. It offers 25+ free models with about 50 requests per day, rising to 1,000 per day once you have bought at least 10 dollars of credits.
Contractual SLAs are only offered on the Enterprise plan.
EU and US in-region routing is available on Business and Enterprise.
Standard and Business have no minimum and no lock-in. Enterprise is sold by contract with invoicing.
Similar Tools
See all alternatives
Amazon's public cloud platform, with over 200 pay-as-you-go services for compute, storage, databases, networking and AI, used by startups through large enterprises and governments.

Portkey is an AI gateway and LLMOps control plane that routes, monitors, and governs requests to 1,600+ LLMs for engineering teams and enterprises.

Northflank is a Kubernetes-based developer platform for deploying services, jobs, databases and GPU workloads on its managed cloud or on customers' own cloud accounts.
Ranked near OpenRouter in AI Software
See OpenRouter in the listOpenRouter is #102 in AI Software by our score.





