OpenRouter
Also known as: OpenRouter.ai
Unified API and marketplace routing agents to 400 plus models through one OpenAI compatible endpoint, with failover, cost controls, and pass through pricing.
OpenRouter is a unified API and marketplace that gives developers and enterprises access to more than 400 large language models from providers including OpenAI, Anthropic, Google, xAI, and DeepSeek through a single OpenAI compatible endpoint. Instead of integrating a separate SDK for each provider, an application points at one API and reaches any model with the same request format, while OpenRouter normalizes responses so the output shape stays consistent regardless of which lab served the request. Founded in 2023 by Alex Atallah, the company is independent, headquartered in New York, and raised a $113 million Series B in May 2026 led by CapitalG at a valuation near $1.3 billion.
The platform is a smart routing layer that selects an endpoint based on price, latency, uptime, and throughput, with automatic failover when a provider is rate limited or unavailable and roughly 25 milliseconds of added overhead. Developers can steer routing with variants such as one for the fastest response and one for the cheapest option, and can bring their own provider keys. OpenRouter processes more than 25 trillion tokens a week across more than 8 million developers, and publishes usage and model rankings as open data that have become a widely referenced signal for real world model adoption.
Beyond text, the platform now supports image, audio, speech, transcription, embedding, and video models, along with governance and cost controls aimed at enterprises moving from single model pilots into multi model production. Pricing is consumption based: there is no subscription to start, free models are available, and paid usage is billed as pass through inference spend with a fee of about five percent on credits. This positions OpenRouter as the routing and gateway layer between agents and model providers, handling reliability, cost optimization, and provider breadth so teams never have to commit to a single model.
Vendor details
Canonical URL
https://openrouter.ai
Category
Agent infrastructure
Subcategory
LLM gateway and routing
Funding status
Independent. Founded 2023. Raised about $173 million total across three rounds, including a $113 million Series B in May 2026 led by CapitalG at a valuation near $1.3 billion. Processes more than 25 trillion tokens a week for 8 million plus developers.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
One OpenAI compatible endpoint reaching 400 plus models across OpenAI, Anthropic, Google, xAI, DeepSeek, and more than 60 active providers. Drop in compatibility with the OpenAI SDK, routing variants for cost and latency, bring your own key support, and multimodal coverage across text, image, audio, speech, transcription, embedding, and video.
In practice
Your agent needs the cheapest capable model per call with no downtime. You point at one OpenRouter endpoint, set a routing variant for lowest cost, and it fails over to another provider automatically when one is rate limited.
You are testing five models across three providers and tired of five SDKs. OpenRouter gives you one OpenAI compatible API and a consistent response shape, so you swap models by changing a string.
Finance wants your AI spend under control as usage scales. You route all traffic through OpenRouter, track spend per key, and use quality aware routing to keep production costs predictable.
Sources & related URLs
Capability coverage
5.0 / 14 capabilities · 36%
| Integrations & Tool CallingTool calling passthrough and 60 plus provider integrations, official docs 2026-07-06 | Full |
|---|---|
| Workflow OrchestrationRouting layer, not an orchestrator | Unable to verify |
| Knowledge Grounding & RAGNo native RAG or knowledge grounding | Unable to verify |
| Human Oversight & GuardrailsNo first class guardrails product | Unable to verify |
| Security, Identity & GovernanceAPI keys and org controls with enterprise governance emerging, official docs 2026-07-06 | Partial |
| Observability & AuditabilitySpend tracking and open model rankings, limited request tracing, official docs 2026-07-06 | Partial |
| Memory & State PersistenceStateless routing, no memory layer | Unable to verify |
| Deployment & Data ResidencySaaS only with limited residency controls, no self host | Partial |
| Prebuilt Agents, Templates & PacksNo prebuilt agents or templates | Unable to verify |
| Triggers & Channel CoverageNo event triggers or channel coverage | Unable to verify |
| Model Flexibility & RoutingRouting across 400 plus models is the core product, official docs 2026-07-06 | Full |
| APIs, SDKs & MCP ExtensibilityOpenAI compatible API and SDK, official docs 2026-07-06 | Full |
| Testing, Debugging & OptimizationModel rankings and comparison, not a full eval suite | Partial |
| Browser & Computer UseNo browser or computer use | Unable to verify |
Pricing
Free to start, pay as you go, about five percent fee on inference credits
usage credits
Included quota
No subscription. You add credits and pay pass through inference spend at provider rates plus a fee of about five percent on credit purchases. Free models are available. Bring your own provider keys is supported.
What is public
OpenRouter publishes its consumption model clearly: no subscription, pass through provider pricing, a fee of about five percent on credit purchases, published per model rates, and free tier models.
Billing mechanics
Prepaid credits are drawn down as requests route to providers. Each request is billed at the serving provider's rate plus the platform fee. Routing variants let you optimize for cost or latency; bring your own key shifts provider billing to your own accounts.
Cost watchouts
Cost tracks inference volume with no ceiling, so agentic workloads that consume many tokens per session can escalate quickly. The roughly five percent fee is on top of already variable provider rates.
Variable cost rationale
There is no fixed plan; cost is entirely a function of inference volume routed through the platform, so spend scales directly with token usage.
Overage / add-ons
Usage is metered continuously against prepaid credits at provider rates plus the platform fee; there is no fixed included quota.
Sales call required
No — self-serve available
Free / trial
Free to start, no subscription; free models available
Lowest paid plan
Pay as you go inference credits with a fee of about five percent; no fixed plan
Commercial notes
Independent, New York based, founded 2023. Raised about $173 million total including a $113 million Series B in May 2026 led by CapitalG at a valuation near $1.3 billion. Processes more than 25 trillion tokens a week across 8 million plus developers.
Related vendors
- AgentOps — Agent observability and reliability platform with broad model and…
- Agno — High-performance agent runtime and framework (formerly Phidata) with…
- Apify — Cloud platform for web scraping and automation with 45,000+ prebuilt…
- Arcade — Authenticated tool calling platform and MCP runtime that handles…
- Arize AI — AI observability and evaluation platform that traces, evaluates, and…
- Beam — Open source serverless sandboxes and GPU cloud for production…