Hi Product Hunt, I'm Saran from the Alye team. We built Alye after watching our own LLM bill grow every month with no answer to a simple question: which feature, model or customer is spending it? Provider invoices give one number. Alye records every model call where it happens and shows: • Cost by service, model and customer, priced for the region and cloud marketplace it ran in (Azure OpenAI, Vertex AI, Bedrock) • Traces of each request, with retries and errors inside • Alerts on spend, monthly forecast, error rate, latency and services that stop reporting, sent to Slack, Teams or email Setup is pip install alye or npm install alye and two lines per service. Nothing sits in your request path, and prompts aren't sent unless you turn it on. It's free for 500,000 calls a month. I'd love to hear what you track today and what's missing. I'll be here all day answering questions.

Alye

See what every LLM call costs, and when one fails

See what every LLM call costs, and when one fails