OpenRouter

What OpenRouter does

OpenRouter is a unified LLM inference routing platform founded in 2023 by Alex Atallah (former co-founder and CTO of OpenSea) and headquartered in New York. The company operates a single OpenAI-compatible API endpoint (https://openrouter.ai/api/v1) that routes requests across more than 400 AI models from over 70 upstream providers — including Anthropic, OpenAI, Google, Meta, xAI, Mistral, DeepSeek, Qwen, Xiaomi, Poolside.ai, Arcee, Nous Research, and others. As of mid-2026, the platform processes approximately 100 trillion tokens per month (25 trillion tokens per week) for more than 10 million global users and serves as the inference layer for over 250,000 downstream applications.

The platform's technical architecture is differentiated by several proprietary components: a normalized reasoning token interface that abstracts provider-specific implementations (Anthropic extended thinking, OpenAI o-series reasoning, Gemini thinkingLevel) into a unified parameter surface; configurable routing modes (Balanced, Nitro, Exacto, Auto Exacto) that let callers trade off cost, latency, and determinism; provider failover and model fallback for uptime; zero data retention (ZDR) enforcement and Sovereign AI data residency routing; prompt caching yielding 60-80% effective price reductions on repeated context; and the Fusion Router, which combines outputs from multiple participant models through a judge model to deliver frontier-tier performance at roughly half the per-token cost. Adjacent modules include Broadcast (auto-instrumented observability to 17+ backends), Server Tools (Web Search, Web Fetch, Image Generation, Advisor, Subagent), and Router variants (Auto, Body Builder, Pareto, Free Models). The platform is deeply embedded in the developer ecosystem via native SDK integrations with Vercel AI SDK, LangChain, Anthropic Agent SDK, Apple Xcode, Replit Agent, LiveKit, Mastra, TanStack AI, PydanticAI, Effect AI, Zapier, and others.

OpenRouter's revenue model combines three streams: (1) usage-based token pricing via prepaid credit purchases — the primary stream, with annualized revenue exceeding $50M as of April 2026, up from $10M+ in October 2025; (2) a small platform/routing fee on BYOK (Bring Your Own Key) requests where customers route existing provider keys through OpenRouter; and (3) recurring enterprise subscription revenue tied to the Enterprise tier featuring workspaces, centralized billing, custom data policies, ZDR enforcement, and organizational budget controls. Go-to-market is multi-modal: API-first and product-led growth via self-serve sign-up, with free BYOK and free model tiers funneling long-tail developers; community-led growth via the annual State of AI report (built on 100T+ tokens of usage data co-published with Andreessen Horowitz); channel partners via SDK ecosystem integrations; and emerging enterprise field sales funded by the $113M Series B closed in May 2026 at a $1.3B post-money valuation led by CapitalG.

Name

OpenRouter

Legal name

OpenRouter, Inc.

Website

https://openrouter.ai

Company type

Private

Founded year

2023

Operating status

Operating

Headcount range

11–50 employees

Short description

OpenRouter operates a unified OpenAI-compatible API that routes inference requests across 400+ AI models from 70+ providers, serving 10M+ developers and 250k+ applications that consume 100 trillion tokens monthly via credit-based usage pricing.

Ownership category

akta.pro rank

Product category

AI Infrastructure / LLM API Gateway

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518210)

SIC

Services-Prepackaged Software (7372), Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

Developer APIs & Node Access (RPC providers, node infrastructure APIs) (FSAPABAC)

Keywords

• LLM API gateway
• AI model routing
• Multi-model inference
• AI infrastructure platform
• Unified AI API

Headquarters

HQ city

New York

HQ country

United States

HQ region

North America

Offices1 record

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Personnel, Marketing or Sales, Operations

Revenue model

1. Token-based API revenue (credit purchases): Primary revenue stream: developers buy credits via OpenRouter, then consume them across 400+ models at per-token rates. Annualized revenue reached $50M+ as of April 2026, up from $10M+ in October 2025 — a 5x+ increase in roughly six months. The credit model operates like prepaid usage-based billing.
2. BYOK platform fee: Secondary revenue stream: when customers bring their own provider API keys (OpenAI, Anthropic, etc.), OpenRouter charges a small platform/routing fee on those requests. This complements direct credit purchases by letting customers route volume while paying providers directly.
3. Enterprise tier revenue: Enterprise customers with centralized billing, workspaces, custom data policies, ZDR enforcement, and organizational budget controls — revenue expanded as part of the Series B fundraise to scale enterprise features.

Pricing tiers

Go-to-market motion5 records

Distribution channels7 records

Marketing channels9 records

Core offering

OpenRouter operates a unified, OpenAI-compatible API gateway at https://openrouter.ai/api/v1 that routes requests across 400+ AI models from 70+ upstream providers (Anthropic, OpenAI, Google, Meta, Mistral, DeepSeek, xAI, Cohere, etc.). Customers buy credits (prepaid usage-based billing) or bring their own provider keys (BYOK) and consume tokens via a single drop-in endpoint that handles provider failover, normalized reasoning tokens, prompt caching, custom data policies, and zero-data-retention enforcement.

Product overview

OpenRouter is a unified LLM infrastructure platform (single product architecture with a modular feature surface) that provides a single OpenAI-compatible API endpoint for accessing 400+ AI models from 70+ providers. The core product, OpenRouter itself, is a routing gateway handling 100T monthly tokens for 10M+ users, with attached modules including Fusion (multi-model combination), Broadcast (auto-instrumented observability to 17+ destinations), Server Tools (Web Search, Web Fetch, Image Generation, Advisor, Subagent, etc.), Plugins (Response Healing, Web Search), and specialized Routers (Auto, Body Builder, Free Models, Pareto, Fusion Router). Supporting infrastructure includes the Enterprise tier for governance/budget controls, the Labs program for experimental features (such as the AI Governance Checklist), Chat for browser-based interaction, Rankings for token consumption leaderboards, Apps for a marketplace of 250k+ built applications, and security/compliance features like ZDR (Zero Data Retention), Sovereign AI for data residency, and Guardrails for prompt injection detection.

Differentiator

Problem solved

Functional benefit

Brands

• Fusion: OpenRouter's multi-model inference product that combines outputs from multiple participant models through a judge model to deliver frontier-level AI performance at roughly half the per-token price.

Products and services

• OpenRouter (Core Unified Interface) The core product is a unified LLM API gateway providing a single OpenAI-compatible endpoint (https://openrouter.ai/api/v1) that routes requests across 400+ AI models from 70+ providers. It serves 10M+ global users, processes 100 trillion tokens monthly, and includes model fallbacks, provider failover, prompt caching, and custom data policies for governance.
• Fusion Multi-model combination product that synthesizes outputs from multiple participant models through a judge model, achieving 64.7% on key benchmarks versus Claude Fable 5's 65.3% at approximately half the per-token cost. Positioned as a cost-efficient alternative to single-model frontier performance.
• Chat

Named customers4 records

Segments4 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration31 records

AI capability15 records

Feature4 records

Partnerships

20 partnerships are on record, tiered flagship technology/payment partner, flagship technology/model partner, strategic infrastructure partner, strategic model partner, flagship distribution partnership, flagship distribution/integration partner, strategic platform integration, flagship sdk integration partner, strategic observability partner, strategic no-code distribution partner, strategic security/credentials partner, distribution partnership for distributed inference, industry standards partner, feature extension partner, strategic observability integration partner, strategic model provider partner and strategic distribution partner.

• Stripeflagship technology/payment partnerTechnology or IntegrationStripe powers OpenRouter's global payment infrastructure including Invoicing, Tax, Radar for Fraud, and Adaptive Pricing to support credit purchases across 150 countries. Every company on the Forbes AI 50 that monetizes does so on Stripe's platform — Stripe additionally acquired Metronome to address AI billing infrastructure. OpenRouter is also integrated with Stripe's AI billing product (billing feature that auto-passes through and marks up AI token costs).
• Anthropicflagship technology/model partnerTechnology or IntegrationOpenRouter provides an 'Anthropic Skin' API that allows Claude Code and the Anthropic Agent SDK to use OpenRouter as their API endpoint via ANTHROPIC_BASE_URL configuration. Hosts 34 Anthropic models including Claude Fable 5, Opus 4.8, Sonnet 4.6, Haiku 4.5 — enabling provider failover and budget controls for Anthropic model usage through OpenRouter.
• Sentient Labs (Arena)strategic infrastructure partnerTechnology or IntegrationOpenRouter is one of the infrastructure partners supplying inference compute for Sentient's Arena initiative — a production-grade benchmarking environment for evaluating AI agents in enterprise workflows alongside Fireworks. Launched February 2026 with Founders Fund, Pantera, and Franklin Templeton as founding participants.
• Poolside.aistrategic model partnerTechnology or IntegrationOpenRouter hosts Poolside.ai's first public foundation models — Laguna M.1 (225B MoE, 23B active) and Laguna XS.2 (33B MoE, 3B active) — built for agentic coding and long-horizon workloads, with free promotional access during launch period.
• Anthropic (Claude Code)flagship distribution partnershipOEM/ Whitelabel/ Licensing PartnerOpenRouter integrates with Anthropic's Claude Code CLI, allowing developers to route their agentic coding workflows through OpenRouter's API and access 320+ LLMs as alternatives to Anthropic's native models — achieved by setting ANTHROPIC_BASE_URL to OpenRouter's endpoint.
• Replitflagship distribution/integration partnerTechnology or IntegrationReplit Agent and Replit Apps integrate OpenRouter as a model provider — developers can bring their own OpenRouter API key into Replit Agent (bypassing Replit-managed credentials) or store it as a project Secret for direct API calls. Replit is also a featured app on OpenRouter's homepage.
• Apple (Xcode)strategic platform integrationTechnology or IntegrationOpenRouter integrates with Apple Intelligence in Xcode 26 (beta) as a chat provider — developers can access 400+ models from Anthropic, Google, Meta, OpenAI etc. directly in Xcode's intelligence pane for code completion, explanation, refactoring, and documentation generation.
• Vercel (AI SDK)flagship sdk integration partnerTechnology or IntegrationOpenRouter provides the official @openrouter/ai-sdk-provider for Vercel AI SDK, enabling Next.js developers to stream text, generate video, and use tools through OpenRouter models via the standard Vercel AI SDK APIs.
• LangChainflagship sdk integration partnerTechnology or IntegrationOpenRouter ships dedicated ChatOpenRouter integration packages for both Python (langchain-openrouter on PyPI) and JavaScript (@langchain/openrouter on npm) — official LangChain integrations supporting streaming, tool calling, structured output, reasoning, multimodal inputs, and provider routing.
• Grafana Cloudstrategic observability partnerTechnology or IntegrationOpenRouter Broadcast auto-generates traces for each API request and broadcasts them to Grafana Cloud, allowing teams to monitor costs, latency, errors, and usage patterns without additional coding.
• Zapierstrategic no-code distribution partnerTechnology or IntegrationOpenRouter is available as a Zapier integration, allowing non-developers to compose AI-powered Zaps across 8,000+ apps using any of 400+ OpenRouter models — including streaming responses, function calling, and multimodal capabilities.
• Infisicalstrategic security/credentials partnerTechnology or IntegrationInfisical integrates with OpenRouter to provide automated API key rotation with zero-downtime credential swaps — uses OpenRouter's Management API keys to provision and rotate regular OpenRouter API keys on a schedule.
• Darkbloom / Eigen Labsdistribution partnership for distributed inferenceTechnology or IntegrationDarkbloom (Eigen Labs' distributed AI computing network) launched on OpenRouter in June 2026, enabling any Mac user to contribute idle compute and surfacing models like gpt-oss-20b and Gemma 426B on OpenRouter with related compute freely available to users.
• OpenAI (Open Responses specification)industry standards partnerStrategic or Co-development PartnerOpenRouter is an early adopter of OpenAI's Open Responses specification — an open standard for agentic AI workflows that enables developers to switch between proprietary and open-source models without rewriting integration code. Other adopters include Hugging Face, Vercel, LM Studio, Ollama, and vLLM.
• Backboard.iofeature extension partnerTechnology or IntegrationBackboard.io launched Bring Your Own Key (BYOK) functionality supporting OpenRouter — enables OpenRouter users to connect their existing keys and add persistent memory, stateful threads, session continuity, and integrated RAG capabilities with zero disruption to existing API routing.
• Elasticstrategic observability integration partnerTechnology or IntegrationElastic integrates OpenRouter within its Agent Builder and new inference pipeline — providing unified access to 500+ models via OpenRouter alongside AI connectors and monitoring tools within Elastic Cloud.
• Mistral AIstrategic model provider partnerTechnology or IntegrationMistral AI's models (including the Mistral 3 Large family under Apache 2.0) are accessible via OpenRouter's unified API, contributing to the 70+ provider catalog.
• Xiaomistrategic model provider partnerTechnology or IntegrationXiaomi's MiMo-V2-Pro (1-trillion-parameter flagship) launched on OpenRouter via the 'Hunter Alpha' stealth release in March 2026 and quickly topped OpenRouter's leaderboard — partnering with five major agent frameworks including OpenClaw to offer free access to developers worldwide.
• Hugging Facestrategic distribution partnerTechnology or IntegrationModels available on Hugging Face (including NVIDIA's Nemotron family) are also surfaced through OpenRouter as part of an open-model distribution loop. Co-adopter of OpenAI's Open Responses specification.
• Google (Gemini)strategic model partnerTechnology or IntegrationGoogle's Gemini models (including Gemini 3.1 Pro Preview and Gemini 3 Flash Preview) are available via OpenRouter. Gemini 2.5 Flash Image was launched with OpenRouter.ai and fal.ai as distribution partners at $30 per million output tokens.

Scale indicators12 records

Recent moves6 records

Expansion highlights7 records

Emerging players

• Together AI: Together AI provides open-model inference with multi-model routing and is listed as an OpenRouter upstream provider. It is an emerging player that overlaps with OpenRouter's catalog and inference routing, and competes for AI-native developer customers.
• Fireworks AI: Fireworks AI offers hosted inference for open and proprietary models with fine-tuning and routing features. It is also partnered with OpenRouter as an upstream provider and Sentient Arena partner, making it an emerging peer at the model-routing/inference layer.

Direct peers

• Helicone: Helicone provides LLM observability, request routing, and provider failover — a direct competitor to OpenRouter's Broadcast observability and routing features. It serves the same developer-first AI customer base with overlapping product capabilities.
• LiteLLM: LiteLLM is an open-source LLM gateway and proxy that standardizes calls to 100+ LLM APIs. It is a direct peer offering similar unified-API and routing functionality, and competes with OpenRouter at the developer-self-serve end of the market.
• LangSmith (LangChain): LangSmith offers LLM tracing, evaluation, and routing within the LangChain ecosystem. It is a direct peer at the observability and governance layer for multi-model AI applications, though bundled with LangChain's broader framework suite.
• Portkey: Portkey is an LLM gateway focused on BYOK governance, observability, and provider routing for production AI deployments. OpenRouter's own blog explicitly compares against Portkey, making it the closest direct peer — same AI gateway category, overlapping enterprise GTM, competing on governance and observability features.

Broad incumbents

• Cloudflare AI Gateway: Cloudflare's AI Gateway offers unified routing, caching, and observability across multiple LLM providers, integrated into Cloudflare's edge network. It is a broad incumbent play — overlapping routing functionality bundled with a much larger CDN/security platform — competing for the same enterprise multi-model workloads.
• AWS Bedrock: AWS Bedrock provides multi-model access (Anthropic, Meta, Mistral, Cohere, Stability) through a single AWS API, with IAM, governance, and enterprise billing. It is a broad hyperscaler incumbent competing for the same enterprise multi-model workloads but as part of the wider AWS portfolio.
• Azure AI Foundry: Microsoft's Azure AI Foundry (formerly Azure AI Studio) offers multi-model routing and governance across OpenAI, Anthropic, Meta, and Mistral models within the Azure cloud. It competes for enterprise governance and multi-model deployments as part of the broader Azure data and AI stack.
• Google Vertex AI: Google Vertex AI's Model Garden exposes multiple foundation models (Google, Anthropic, Meta, Mistral) through a unified endpoint with enterprise governance. It is a broad hyperscaler incumbent whose multi-model offering directly overlaps OpenRouter's enterprise value proposition.

Market position

Strengths5 records

Weaknesses4 records

Competitive moat6 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/openrouterai
• GitHubgithub.com/OpenRouterTeam
• YouTubeyoutube.com/@OpenRouterAI
• LinkedInlinkedin.com/company/openrouter

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Funding overview

Funding rounds3 records

Investors12 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about OpenRouter