Put CI agents on a separate model budget.
Review bots, test generators, and release jobs should not share the same model key as production. ProxyLLM gives GitHub Actions its own capped key, and OpenAI-model calls bill to your ChatGPT subscription through Codex Hosted.
$129/month SaaS. Bring your own model keys. No inference markup.
Three steps to connect.
Create a CI key
Generate a ProxyLLM key just for GitHub Actions and store it as PROXYLLM_API_KEY in repository or organization secrets.
Set the base URL in the job
Export OPENAI_BASE_URL as https://api.proxyllm.ai/v1 and pass the ProxyLLM key to your tests, review bot, or release job.
Cap unattended runs
CI runs on every push and retries what fails. ProxyLLM keeps that automation under its own cap, with per-key request logs and subscription billing instead of a meter.
Secrets in, gateway out.
Most CI AI tools read OpenAI-compatible environment variables already.
env:
OPENAI_BASE_URL: https://api.proxyllm.ai/v1
OPENAI_API_KEY: ${{ secrets.PROXYLLM_API_KEY }}
steps:
- uses: actions/checkout@v4
- run: npm testRun your AI workloads on your ChatGPT subscription.
ProxyLLM runs OpenAI's Codex for you, signed in with your own ChatGPT account. Your apps call one OpenAI-compatible endpoint and the work bills to your flat plan instead of per-token API pricing.
Make unattended AI jobs accountable.
One capped key for CI, a request log for every run, flat billing through Codex Hosted. No pull request gets an unlimited model tab.