Hey Product Hunt! 👋 I built Troxen after repeatedly watching developers and startups lose massive amounts of money to uncapped AI API bills. Connecting directly to providers like OpenAI or Anthropic leaves your credit card exposed to unexpected traffic spikes and runaway scripts. On top of that, handling rate limiting and caching in-house is frustrating—custom setups frequently get wiped out, misconfigured, or end up triggering constant API errors for end-users. I wanted to build a reliable, drop-in safety net that handles the heavy lifting without forcing developers to constantly rewrite custom gateway logic. Troxen is a drop-in API proxy gateway built explicitly for AI workloads: • Hard Per-User Budgets: Set strict spend caps per user ID so runaway usage gets blocked before provider calls happen. • Semantic Caching: Cache repeat prompts across your user base so identical requests cost $0 and avoid redundant provider calls. • 1-Line Integration: Simply swap the base URL in your SDK to proxy.troxen.dev/v1. We are officially opening up the waitlist for our private beta today! I’d love to get your feedback, answer any technical questions, and hear what features you need most in an API gateway.

Troxen | Put a ceiling on your LLM spend

AI API gateway for hard per-user budgets & caching

AI API gateway for hard per-user budgets & caching