OpenRouter raises $40M in funding for its one-stop shop marketplace for all AI models

OpenRouter, an AI startup building a unified API layer for AI inference, has secured $40 million in combined Seed and Series A funding to support developers struggling to keep up with the surge of new AI models.

The round was led by Andreessen Horowitz and Menlo Ventures, with backing from Sequoia and a group of well-known angels. OpenRouter, now valued at around $500 million, helps route AI prompts to the most efficient large language models based on speed, cost, and other variables.

OpenRouter’s $40M Raise Signals Shift Toward Unified AI Infrastructure

The company is betting on a simple idea: AI stacks are getting messier, and someone needs to make them less painful to use. With hundreds of large language models (LLMs) popping up—from OpenAI to Mistral to Google—developers and enterprises are juggling different APIs, contracts, and quirks. OpenRouter’s pitch is that you can plug into a single API and access over 400 models without rewriting your code.

Founded in 2023 by Alex Atallah (CEO) and Louis Vichy, OpenRouter gives developers a single point of access to major LLMs, with unified billing, token accounting, and real-time spend tracking. It’s already become a go-to for AI builders and is seeing fast adoption across both startups and large companies.

“There’s been a Cambrian explosion of models,” said Atallah,  who previously co-founded OpenSea, an online marketplace for non-fungible tokens. “Our business is a one-stop shop for all models,” Atallah told the Wall Street Journal.

“We’re seeing more teams hitting the same walls,” said Atallah. “They’ve cobbled together internal solutions, but those fall apart at scale. So they’re ripping them out and replacing them with OpenRouter. It lets them stop worrying about the plumbing and focus on what actually matters to their business.”

The new capital will help the company move faster on product development, add support for more models, and scale up its enterprise tools.

Here’s what the traction looks like so far:

• OpenRouter’s inference volume has skyrocketed. Back in October 2024, it was running at a $10 million annual rate. As of May 2025, that number has jumped to more than $100 million.

• The developer community is growing fast. Over a million developers have used OpenRouter’s API since launch.

• The infrastructure is already handling a serious load. It routes billions of requests and trillions of tokens weekly, with failover across 50+ cloud providers and global edge deployment averaging ~25ms of overhead.

• Big names are already plugged in. OpenRouter integrates with VS Code, Zapier, Cloudflare, Make.com, n8n, and PostHog, among others.

• The AI labs are watching closely. OpenRouter partnered with OpenAI on the quiet rollout of GPT-4.1, giving its users early access to the model and sharing back valuable real-world feedback with the lab.

The enterprise version of OpenRouter includes zero-logging as the default, flexible routing based on data policies, unified billing, and support for companies bringing their own capacity. Whether a startup is spending $500/month or a global platform is pushing millions through inference costs, the company says it can handle the load.

Andreessen Horowitz’s Anjney Midha calls it “the kind of infrastructure play that defines new categories.” Menlo Ventures’ Matt Murphy pointed to the shift in developer behavior. “We’re seeing people actively tuning performance across models and providers. That’s driving a surge in OpenRouter usage.”

Tabarak Khan from OpenAI added, “Their diverse and active developer community has shared valuable feedback on how our models perform in practice. We’re excited to keep building alongside them.”

With the space heating up, OpenRouter is clearly positioning itself as the glue between AI labs, developers, and enterprise buyers—one unified API to make all these moving parts talk to each other.