OpenRouter
Also known as: OpenRouter.ai
Unified API and marketplace routing agents to 500+ models from 80+ providers through one OpenAI compatible endpoint, with failover, spend controls, in-region routing and the Ori agent CLI.
OpenRouter is a unified API and marketplace that gives developers and enterprises access to more than 500 AI models from more than 80 providers, including OpenAI, Anthropic, Google, xAI and DeepSeek, through one OpenAI compatible endpoint. An application points at one API and reaches any model with the same request format, and OpenRouter load balances across providers, fails over when one is down or rate limited, and lets the customer steer routing by price, latency, throughput, provider and data policy. It covers text, image, audio, speech, transcription, embedding and video models.
Around the routing sit controls for teams. Workspaces separate keys, routing defaults and observability per team or environment, and guardrails cap spend and restrict models and providers per member or key. Zero data retention routing sends requests only to providers that keep no data, and in-region routing keeps prompts and completions inside the EU or the US on Business and Enterprise plans.
Traces broadcast to observability tools such as Langfuse, Datadog and LangSmith, custom classifiers tag each request, and Enterprise adds SSO and SCIM. For builders, the Agent SDK runs agent loops with tools, stop conditions and human approval gates. Beta server tools let OpenRouter run web search, web fetch, shell commands and subagents during a request. Ori, OpenRouter's agent CLI, runs existing agent CLIs on OpenRouter, schedules headless agent runs and evaluates agents across models.
Pricing is usage based. A free tier offers free models, and pay as you go usage is billed at provider rates plus a platform fee of 5.5 percent on Standard or 8 percent on Business, with Enterprise discounts through sales. In August 2026 OpenRouter announced that it is joining Stripe, subject to closing, and said it will keep operating under the same name, product and roadmap.
Vendor details
Canonical URL
https://openrouter.ai
Category
Agent infrastructure
Subcategory
LLM gateway and routing
Funding status
OpenRouter, Inc. On 19 August 2026 OpenRouter announced that it is joining Stripe; the transaction is subject to customary closing conditions, and Stripe's newsroom lists it as an agreement to acquire. OpenRouter says it will keep operating under the same name, product and roadmap.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
One OpenAI compatible endpoint reaches 500+ models from 80+ providers, with provider routing, fallbacks and variants, bring your own keys, and client SDKs for TypeScript, Python and Go plus an Agent SDK. Server tools cover web search, web fetch, shell and subagents. Traces broadcast to Arize AX, Braintrust, Datadog, Langfuse, LangSmith, Grafana, OpenTelemetry, S3 and webhooks. SSO works through Okta, Entra ID, Google Workspace or SAML, with SCIM group mappings.
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
Agentic Index coverage score
10.5 / 14 capabilities · 75%
| Integrations & Tool Calling | Full |
|---|---|
|
Custom tools work across the catalog: tool calling uses one request format for any tool capable model, Auto Exacto routes tool calling requests to the providers that handle them best, and the Agent SDK's tool() helper defines a customer's tools and executes them in the agent loop, including tools from MCP servers. An agent built on OpenRouter takes authenticated action in outside systems through the customer's own tools. Sourceopenrouter.ai/docs/guides/features/tool-callingread 2026-09-22 |
|
| Workflow Orchestration | Partial |
|
The Agent SDK runs multi turn agent loops, calling tools and deciding the next step until a stop condition is met (step count, a specific tool call, maximum cost), and the subagent server tool lets a model hand self contained tasks to a worker model, optionally with its own tools, mid generation. There is no workflow definition, deterministic node or versioned flow, so a workflow cannot mix deterministic nodes with agent steps. Sourceopenrouter.ai/docs/agent-sdk/overviewread 2026-09-22 |
|
| Knowledge Grounding & RAG | Not documented |
|
OpenRouter routes embedding requests to embedding models, and its RAG cookbook shows the customer building and storing their own index with those embeddings. The web search and web fetch server tools reach the public web, which is not the customer's corpus. OpenRouter maintains no retrieval structure over the customer's own documents, and routing embedding calls does not create one. Sourceopenrouter.ai/docs/api_reference/embeddingsread 2026-09-22 |
|
| Human Oversight & Guardrails | Full |
|
The Agent SDK (@openrouter/agent) includes an approval gate. A tool marked requireApproval pauses execution when the model calls it, so a person can approve or reject each call before it runs, and a StateAccessor carries approval decisions across separate request cycles, for example in a web application. Workspace guardrails also cap spend and restrict models and providers per member or key. Sourceopenrouter.ai/docs/agent-sdk/call-model/tool-approval-stateread 2026-09-22 |
|
| Security, Identity & Governance | Full |
|
Enterprise plans get self serve SSO through Okta, Microsoft Entra ID, Google Workspace or any SAML provider, and SCIM group mappings grant workspace access from identity provider groups, with an audit log of mapping and membership changes. Workspaces isolate API keys, routing defaults and guardrails per team, and guardrails per member and per key restrict models, providers and spend. OpenRouter describes itself as SOC 2 compliant and GDPR compatible without naming a report, auditor or date. Sourceopenrouter.ai/docs/guides/features/ssoread 2026-09-22 |
|
| Observability & Auditability | Full |
|
Every plan carries activity logs and export. Broadcast sends trace data for every API request, without instrumentation in the customer's code, to Arize AX, Braintrust, Datadog, Langfuse, LangSmith, Grafana, an OpenTelemetry collector, S3, a webhook and other destinations, configured per workspace by an organization admin. Custom classifiers tag each generation with dimensions the customer defines, and the tags appear in the logs and roll up in the activity view, while SCIM mapping changes keep a separate audit log. Together these give step by step inspection of what the agent sent and received, audit records separate from runtime traces, and export to the customer's own tools. Sourceopenrouter.ai/docs/guides/features/broadcastread 2026-09-22 |
|
| Memory & State Persistence | Partial |
|
The Agent SDK's StateAccessor persists conversation state (message history, tool results and approval decisions) between callModel invocations, and a hosted intern continues the same run when a later prompt carries its session_id. That is conversation state carried across turns of one conversation, and there is no memory layer with its own scope and lifetime. Sourceopenrouter.ai/docs/agent-sdk/call-model/tool-approval-stateread 2026-09-22 |
|
| Deployment & Data Residency | Full |
|
In-Region Routing lets a customer send requests through a region specific base URL (eu.openrouter.ai or us.openrouter.ai). The request is decrypted inside that region and routed only to provider endpoints there, and prompts and completions never leave it. It is available on the Business and Enterprise plans. Zero Data Retention routing also sends requests only to providers that do not retain data. Sourceopenrouter.ai/docs/guides/features/in-region-routingread 2026-09-22 |
|
| Prebuilt Agents, Templates & Packs | Not documented |
|
Ori Harness starts agent CLIs the customer already uses, such as Claude Code, Codex and OpenCode, with OpenRouter credentials and models, and those agents are other vendors' products. Presets are saved request configurations, and Ori's built in skills set up and run Ori itself. OpenRouter has no ready made workflows, templates or role specific agents of its own. Sourceopenrouter.ai/docs/guides/ori/harnessread 2026-09-22 |
|
| Triggers & Channel Coverage | Full |
|
The Ori runtime provides a built-in cron scheduler. Features declare schedules, several per feature, that start headless runs with a durable event log, an overlap policy and optional jitter. Schedules can catch up on runs missed during a restart and can be disabled without deletion, and ori schedules reports whether each timer is armed. Those schedules start agent runs with no person initiating them. Sourceopenrouter.ai/docs/guides/ori/changelogread 2026-09-22 |
|
| Model Flexibility & Routing | Full |
|
Model and provider choice sit with the customer. OpenRouter offers 500+ models from 80+ providers behind one API, and a provider object in each request sets order, allowed and ignored providers, price, latency and throughput sorting and data policies, on top of default load balancing, model fallbacks, variants such as :nitro and :floor, and an auto router. Sourceopenrouter.ai/docs/guides/routing/provider-selectionread 2026-09-22 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
OpenRouter offers a REST API with a full API reference, client SDKs for TypeScript, Python and Go, the @openrouter/agent Agent SDK, OAuth PKCE for apps that sign users in, Management API keys for programmatic key and spend control, and an Analytics API. Sourceopenrouter.ai/docs/client-sdks/overviewread 2026-09-22 |
|
| Testing, Debugging & Optimization | Full |
|
Ori Eval tests the customer's agent on real prompts from the customer's project. It writes an eval file, runs the models being compared with one harness and one model held fixed for each run so repeated runs use the same configuration, and returns scores and a recommendation, and features can ship evaluations beside their code through the ori eval command. Retries and model fallbacks also sit in the routing layer. Sourceopenrouter.ai/docs/guides/ori/evalread 2026-09-22 |
|
| Browser & Computer Use | Partial |
|
The web fetch server tool lets any model fetch a URL during a request. OpenRouter fetches and extracts the page, using the provider's native fetch where available and otherwise Exa, and returns the text to the model. That is a headless fetch with a third party engine wired in, and there is no hosted browser, desktop session or remote computer control that an agent drives. Sourceopenrouter.ai/docs/guides/features/server-tools/web-fetchread 2026-09-22 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Default provider routing on OpenRouter no longer picks mainly on price. The new router tracks each endpoint's real price, latency, throughput and error rate, chooses the best mix of cost and speed for each request, and moves traffic away from an endpoint while it has trouble. In OpenRouter's measurements, time to first token improved 16.9%, throughput rose 16.3% and capacity errors fell 46%, with costs up 6.0%.
Bears on: Deployment / data residency
View sourcePrivate Endpoints, which Enterprise customers can have switched on through their account representative, let an organization route its own fine tuned models, dedicated deployments or reserved capacity through OpenRouter under an existing catalog model slug. Private endpoints are visible only within the organization, move to the front of its provider list by default, and can overflow to public providers when rate limited or unavailable.
Bears on: Deployment / data residency
View sourceA new Security Center, on every OpenRouter plan, lists all API keys across workspaces, scores each by how much a leaked copy could spend, and lets admins disable keys or add spend limits in bulk. A maximum key lifetime setting rejects keys that never expire, existing ones included.
Bears on: Security / enterprise
View sourcePricing
Pay as you go at provider rates plus a 5.5% platform fee (8% on Business). A free tier is available.
Prepaid usage credits, drawn down per request at the serving provider's rate plus a platform fee.
Included quota
The free tier includes 25+ free models, 4 free providers and 50 requests a day. Paid plans have no subscription. Usage is billed at provider rates plus the plan's platform fee, with rate limits passed through from providers.
What is public
OpenRouter publishes every plan, with Free, Standard at a 5.5% platform fee, Business at 8% and Enterprise with fee discounts through sales, along with per model rates and the bring your own key allowances.
Billing mechanics
Prepaid credits by credit card, crypto and other methods on Standard and Business. Enterprise adds invoiced billing and credit lines. Each request is billed at the serving provider's rate plus the platform fee. Bring your own keys moves provider billing to the customer's own accounts.
Cost watchouts
Cost tracks inference volume with no ceiling, so agent workloads that use many tokens per session can climb quickly. The platform fee (5.5% Standard, 8% Business) sits on top of variable provider rates, and bring your own key traffic is charged 5% above the monthly free allowance.
Variable cost rationale
Tokens are metered. Every request is billed at the serving provider's rate plus a platform fee of 5.5% on Standard or 8% on Business, so spend rises directly with inference volume and with the models chosen. There is no subscription and no ceiling. Credits are prepaid, so usage draws down a balance loaded in advance, and Enterprise negotiates fee discounts. Traffic on the customer's own keys moves provider billing to its own accounts and carries no platform fee up to $25,000 of list price inference a month, then 5%.
Overage / add-ons
Paid plans include no quota. Every request is billed at the serving provider's rate plus the plan's platform fee (5.5% Standard, 8% Business, discounted on Enterprise). Bring your own key traffic is free up to $25,000 of list price inference a month on Standard and Business ($200,000 on Enterprise), then 5%.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free tier with 25+ free models and 50 requests a day, no subscription
Lowest paid plan
Standard, pay as you go at provider rates plus a 5.5% platform fee, with no subscription
Commercial notes
OpenRouter, Inc. announced in August 2026 that it is joining Stripe, subject to customary closing conditions, and says the product, name and roadmap continue unchanged.
Key ambiguities
Enterprise fee discounts are quoted by sales. How the platform fee applies to failed or fallback attempts is answered in the pricing FAQ.
Support SLA / resale
Community support on Free and email support on Standard and Business. Enterprise adds a dedicated support team and contractual SLAs.
Related vendors
- AgentOps — Agent observability and debugging platform: open source SDKs trace…
- AgentPhone — AgentPhone gives the AI agents a team builds their own US or…
- Agno — Python agent framework and AgentOS runtime (formerly Phidata) for…
- AIsa — Resource and payment gateway for AI agents, with one key to 110+…
- Albato — Albato is a no code integration platform that sells an embedded,…
- AlphaBitCore — AI control plane for regulated financial firms: one gateway enforces…
Alternatives to OpenRouter
The closest documented capability profiles to OpenRouter among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Composio9.5 / 14A lighter documented profile than OpenRouter
- TrueFoundry9.5 / 14A lighter documented profile than OpenRouter
- Braintrust10.0 / 14Adds documented Prebuilt Agents, Templates & Packs
- LangWatch10.0 / 14Adds documented Prebuilt Agents, Templates & Packs
- Windmill12.0 / 14Adds documented Prebuilt Agents, Templates & Packs
- Bernstein10.5 / 14Adds documented Prebuilt Agents, Templates & Packs
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded