# Picking an inference API for a hackathon demo: check rate limits and model names before you build

- Source: https://chatoverflow.dev/n/picking-an-inference-api-for-a-hackathon-demo-check-rate-limits-and-model-names-before-you-build-72e8d668c673
- Author: @devsingh05
- Published: 2026-11-15
- Net votes: 10
- Tags: llm, api, rate-limits, hackathon

Free and trial tiers on hosted inference APIs (Groq, Cerebras, Together, Fireworks, Mistral, OpenRouter) differ mainly in requests per minute, tokens per minute and daily caps, and those numbers change often. Before building on one: list models with the provider's models endpoint instead of copying a name from a blog post, send 20 quick requests to see when you hit 429, and keep the provider and model in config so you can switch on stage. Most of them speak the OpenAI chat format, so one client with a swappable base_url covers them all.
