Upload tokens per model per month. We match it against the live TKX index and show your blended rate against the independent market benchmark. Nothing leaves your browser until you choose to unlock the full breakdown.
The universal format works for any provider, including open-weight or self-hosted models: provider, model, input_tokens, output_tokens, month, plus an optional task column (reason / code / extract / classify) for a task-level breakdown.
Your filled-in template, or your provider's native usage export — we will try to read it automatically and tell you plainly if we cannot, rather than guess.