AI inference gateway reports rapid growth across users, traffic and token volume as its research-driven adaptive routing engine scales in production
SAN FRANCISCO and TOKYO, Aug. 18, 2026 /PRNewswire/ — OrcaRouter, the AI inference gateway built to make accessing and routing across AI models faster, cheaper and more open, today announced that it is on track to reach a $10 million annualized revenue run-rate this month, approximately 10 weeks after launch.
The milestone comes as AI routing rapidly emerges as a major infrastructure category. OpenRouter, one of the pioneers of the market, is reportedly being acquired for more than $7 billion roughly 40 months after launch.
OrcaRouter is positioning itself as the next challenger in the category.
In the past four weeks, OrcaRouter has seen:
- 30x growth in new users
- 20x growth in weekly token spend
- 10x growth in search traffic
- A rise from unranked to No. 12 globally on Hugging Face
- Peak capacity of up to 1 million simultaneous routing decisions
Research-driven routing, built for production
At the core of OrcaRouter is its adaptive model-routing engine, designed to determine which model should handle each request rather than sending every workload to the same expensive frontier model.
The underlying approach is described in the research paper “OrcaRouter: A Production-Oriented LLM Router with Hybrid Offline–Online Learning.”
OrcaRouter combines adaptive model selection, cache-aware routing and frontier escalation to route simpler workloads toward more efficient models while escalating difficult requests when higher capability is required.
In production, the system is designed to help customers achieve 40%+ savings in inference spend while retaining approximately 99% of target quality, according to OrcaRouter’s internal benchmarks.
The routing infrastructure has handled peaks of up to 1 million routing decisions simultaneously, demonstrating the system’s ability to apply adaptive routing at production scale.
Free models, adaptive routing and zero-fee BYOK
OrcaRouter’s early growth has centered around three principles:
Free access. OrcaRouter maintains a rotating selection of popular models that developers can access for free.
Smarter routing. Its adaptive routing engine automatically balances model quality, cost and workload requirements rather than treating every request equally.
No lock-in. Developers can bring their existing provider API keys to OrcaRouter through BYOK and manage and route them through one gateway with 0% BYOK fees.
Users can therefore keep their existing provider relationships, credits and pricing while using OrcaRouter as the routing and management layer.
A rapidly expanding AI infrastructure market
The emergence of multi-model applications is changing how developers consume AI inference. Applications increasingly use combinations of frontier, open-source, specialized and low-cost models rather than relying on a single provider.
OrcaRouter is building the infrastructure layer between those applications and the growing global supply of AI inference.
Its long-term goal is to make model inference increasingly dynamic: matching every request with the appropriate model and provider based on quality, price, latency, availability and other constraints in real time.
Media Contact:
OrcaRouter Communications
[email protected]
+1 650-609-7501
orcarouter.ai
View original content to download multimedia:https://www.prnewswire.com/news-releases/openrouter-alternative-orcarouter-on-track-for-10m-annualized-revenue-run-rate-just-10-weeks-after-launch-302852928.html
SOURCE OrcaRouter
