Free
$0
Testing with free models
- 25+ free models
- 4 free providers
- 5 workspaces
- community support
- 50 requests/day
OpenRouter
One API and one bill for hundreds of AI models across many providers
About OpenRouter
OpenRouter is a unified API and marketplace that gives developers access to hundreds of large language models from many providers through a single endpoint, a single API key and a single bill. Its chat endpoint follows the OpenAI request format, so code written for OpenAI SDKs can usually be pointed at OpenRouter by changing the base URL and model name. The company describes the catalogue as 500+ models from 80+ providers, including text, image, video, speech-to-text and text-to-speech models.
Best for
- Switching or comparing models without code changes, adding provider fallbacks, consolidating LLM billing across vendors
- Model Comparison and Switching: Teams test several models from different vendors behind the same API and swap them without rewriting integration code.
- Resilient Production Inference: Applications use automatic fallbacks across providers to keep serving requests during provider outages.
- Consolidated LLM Billing: Organizations pay for many providers through one credit balance with budgets and spend controls.
- Agent and Coding Tool Backends: Agent frameworks and coding agents route requests to different models through OpenRouter's API or hosted MCP server.
- Prototyping on Free Models: Developers experiment with free model variants before committing credits.
Not ideal for
- Non-technical users wanting a finished chat product, teams that need a single flat monthly subscription
Popular with
- Software Developers
- AI Engineers
- Startups
- Product Teams
- Enterprises building LLM applications
Key features
- Unified API
- Automatic Fallbacks
- Auto-Routing and Provider Controls
- Free Models
- Prompt Caching
- Bring Your Own Key
- Multimodal Models
- Rankings and Benchmarks
- Regional Routing and Private Endpoints
- Hosted MCP Server
Pros and cons
Pros
- Single OpenAI-compatible endpoint covering 500+ models from 80+ providers, so switching models rarely needs code changes.
- Inference passes through at provider list prices with no markup and no minimums or lock-in.
- Automatic provider fallbacks reduce exposure to outages and rate limits.
- Free models and a free router allow testing without paying.
- Enterprise options include SSO/SAML, SLAs, EU and US in-region routing and private endpoints.
Cons
- A 5.5% (Standard) or 8% (Business) platform fee applies to every credit purchase, and platform fees are non-refundable.
- The free tier has no auto-routing, budgets, prompt caching or BYOK, and is limited to 50 requests per day.
- No official mobile app or desktop client; mobile access is through third-party clients.
- Business and Enterprise pricing, and the Stripe acquisition announced in August 2026, add some uncertainty about future terms.
Pricing
Plans as listed by the vendor on . Check their site for current prices.
Standard
Most popular5.5%/one-time
Pay-as-you-go with a fee on credit purchases
- 500+ models
- 80+ providers
- auto-routing
- budgets
- prompt caching
- management API keys
- email support
Business
8%/one-time
Pay-as-you-go for larger teams
- Everything in Standard
- 1,000 workspaces
- EU and US in-region routing
- Workload Identity Federation
- advanced notifications
Enterprise
Custom/custom
Contract-based for large organizations
- Everything in Business
- SSO/SAML
- SLAs
- invoicing
- custom rate allowances
- dedicated support team
Our ratings
Scored out of 10 by the Finder Atlas editorial team.
8.3overall
- Ease of Use
- 8.6
- Feature Depth
- 8.5
- Value for Money
- 8.3
- Performance & Speed
- 8.0
- Documentation & Support
- 8.2
Platforms
- Web
- API
OpenRouter alternatives
Other tools people compare with OpenRouter.
- Compare
GreptileFreemiumAI code review that checks every pull request against your full codebase context
- Compare
MomenticFreemiumAI QA platform that writes, runs and maintains end-to-end tests from plain English
- Compare
BrowserbaseFreemiumCloud browser infrastructure and APIs that let AI agents browse and act on the web
- Compare
Browser UseFreemiumOpen-source library and cloud platform for AI agents that control real web browsers
- Compare
GroqFreemiumFast, OpenAI-compatible inference API for open models, running on custom LPU chips
Release history
Private Endpoints
Bring fine-tuned models and capacity to OpenRouter.
ElevenLabs is now available on OpenRouter
ElevenLabs is now available on OpenRouter.
Security Center
Tools to manage and de-risk API keys.
Batch API
Half-price inference by bundling requests.
In-Region Routing
Route inference in the US or EU.
OpenRouter announced it is joining Stripe, subject to customary closing conditions
OpenRouter announced it is joining Stripe, subject to customary closing conditions.
Reviews
4.5
Based on 2 reviews
- Mout Liang2026-11-084 out of 5
OpenRouter gives one OpenAI-compatible API and one bill for hundreds of models across providers. Switching models without changing client code and the automatic provider fallbacks are the main wins; the fee on credit purchases is the trade-off to keep in mind.
- Jiacheng Feng2026-11-085 out of 5
OpenRouter makes it easy to try many LLM providers through one OpenAI-compatible API and one bill. Switching models without rewriting client code is the main win, and free models help for quick tests. The credit fee on purchases is the trade-off to keep in mind.
Frequently asked questions
Does OpenRouter log my prompts and completions?
According to its FAQ, OpenRouter logs basic request metadata such as timestamps, model and token counts, but not prompts or completions. Users can opt in to prompt logging in exchange for a 1% discount.
Does OpenRouter add a markup to model prices?
Inference is billed at each provider's list price with no markup. A platform fee (5.5% on Standard, 8% on Business) applies when credits are purchased.
What are the limits on free models?
Without purchased credits, free usage is limited to 50 requests per day. After buying at least $10 in credits, the limit rises to 1,000 per day. Free models are also capped at 20 requests per minute and are generally not suited to production use.
Can I get a refund on unused credits?
Refunds for unused credits can be requested within 24 hours of purchase from the Credits page. Platform fees are non-refundable and cryptocurrency payments are never refundable.
What happens if a provider goes down?
OpenRouter automatically falls back to the next provider when a provider returns an error, and failed attempts that error are not charged for model tokens.
Can I use my own provider API keys?
Yes. Standard and Business accounts get $25,000 per month of list-price BYOK usage with no OpenRouter fee, and a 5% fee applies above that.
More Developer Tools tools
View all
DeepSeek
AI Assistants
Free AI chat app and low-cost API running DeepSeek's open-weights V4 models
Freemium
Claude Code
Coding AI
Anthropic's coding agent that edits code and runs commands from terminal, IDE or browser
Paid5.0
Cursor
Coding AI
AI coding agent and VS Code-based editor for multi-file edits and cloud agents
Freemium5.0
