Integration documentation

Set up RunLog

1. Create your account

Open Account, create an account with the email used at checkout, confirm your email if requested, and sign in. Your subscriptions and usage come from your authenticated organization. Keys can be created and revoked there; a new key is shown once.

2. Choose your integration

Prompt Pricer provides browser cost estimates. Model catalog prices and token counts are estimates; verify provider billing before making purchasing decisions.

Benchmark sends your prompt to selected models through the configured gateway. It requires an active entitlement and a customer key. Latency, token counts and cost are returned from the real request; unavailable values are not estimated.

Ship-Check source and installation instructions · Loop-Guard source and installation instructions · Prompt Pricer source

Use the README for the published integration and pin a reviewed commit in CI. Do not assume an npm package, PyPI package, Marketplace listing or GitHub App is available unless the repository explicitly documents it.

3. Call the Benchmark API

Store your customer key in a deployment secret named RUNLOG_API_KEY. This example sends a real prompt; replace the model ID with one supported by the configured gateway.

curl https://runlog.cc/api/v1/benchmark \
  -H "Authorization: Bearer $RUNLOG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"Summarize the purpose of a retry limit.","models":["provider/model-id"]}'

The response contains a results array with model, output, latency_ms, prompt_tokens, completion_tokens, cost_usd and error fields. A model failure is shown individually. Authentication, entitlement or configuration failures return an HTTP error.

Need help?

Contact support for billing, missing access or integration issues.