I built this because I was frustrated by the high margins OpenAI and Anthropic take from small developers. It’s hard to build a sustainable AI business when such a huge chunk of your revenue goes directly to API costs.
The Idea: Instead of spending our budget on Google or Meta ads, we’ve decided to use that marketing capital to subsidize token costs for our users. We’re essentially buying growth by paying for 30% of your API bills.
The Tech:
1-Line Migration: It’s 100% OpenAI SDK compatible. You just change the base_url.
Zero-Logging: We built this as a stateless proxy. We don’t store prompts or completions.
Performance: Routed through a global edge network to keep overhead under 20ms.
Stability: We use high-tier enterprise channels to ensure uptime.
We are losing money on every request right now, but we believe that providing the most affordable, high-quality gateway is the best way to earn the trust of the developer community.
I’ll be here all day to answer technical questions about our routing, latency, or how we handle the subsidy model. Would love your feedback!
Why I’m subsidizing LLM tokens for other devs
Hi HN, I’m the creator of HeyToken.
I built this because I was frustrated by the high margins OpenAI and Anthropic take from small developers. It’s hard to build a sustainable AI business when such a huge chunk of your revenue goes directly to API costs.
The Idea: Instead of spending our budget on Google or Meta ads, we’ve decided to use that marketing capital to subsidize token costs for our users. We’re essentially buying growth by paying for 30% of your API bills.
The Tech:
1-Line Migration: It’s 100% OpenAI SDK compatible. You just change the base_url.
Zero-Logging: We built this as a stateless proxy. We don’t store prompts or completions.
Performance: Routed through a global edge network to keep overhead under 20ms.
Stability: We use high-tier enterprise channels to ensure uptime.
We are losing money on every request right now, but we believe that providing the most affordable, high-quality gateway is the best way to earn the trust of the developer community.
I’ll be here all day to answer technical questions about our routing, latency, or how we handle the subsidy model. Would love your feedback!