TokelyRequest IDs, attempt-level logs, token and cost usage, exports, spending limits and signed webhooks for every LLM call through one gateway.
By TokelyReviewed Every Tokely response carries an X-Request-ID that links to a dashboard log with the model, routing mode, attempts, duration, status, tokens and charge. Usage can be exported, budgets are enforced per account and key, and signed webhooks report account events.
Log the X-Request-ID your application receives. In the dashboard it resolves to the request's model, endpoint, routing mode, each supplier attempt with its outcome and duration, the final HTTP status, token counts including cache reads and writes, and the amount charged.
Usage is recorded per request and per API key, including cached-input tokens billed at their own rate. Export usage for your own reporting. Recorded savings compare each charge with the model developer's price at request time where a comparable reference exists.
Daily and monthly account budgets and per-key limits are reserved before a request runs, so an overrun is rejected with a distinct error instead of being billed. Signed webhooks deliver account events to your own systems.
Tokely observes traffic that goes through its gateway. It does not store API prompts or completions — only usage metadata for billing — and it does not run evaluations, trace your application code or ingest logs from other providers. Pair it with a tracing tool if you need full application traces.
For request-level cost, usage, errors and routing, no: the gateway records them. For prompt evaluation or end-to-end application traces, use a dedicated tracing tool alongside Tokely.
Yes. Each request records cache-read and cache-write tokens, and they are billed at the model's cache rates, not again as regular input.
Check the model, compare the rate, and connect with a Tokely key.