What is a Claude token and how does it affect context and cost?

Input tokens represent the context sent to the model

Depending on the integration, input can include system instructions, user and assistant history, retrieved documents, tool schemas and tool results. Long-running agents may carry substantial context across turns, which is why provider-reported usage is more reliable than counting words in one prompt.

Output tokens represent generated content

Generated answers, code and structured tool arguments consume output tokens under the provider billing rules that apply to the selected model. Input and output can have different prices, so cost estimates should keep them separate.

Use estimates for comparison and provider data for accounting

A browser estimator can help compare two pieces of visible text, but the provider is the source of truth for billable usage. Capture reported input, cached input and output for representative tasks before making a budget or optimisation decision.

Measurement checklist

1. Choose a representative completed task, not an artificial one-line prompt.
2. Record the selected model, provider input, cached input, output, retries and final result.
3. Change one optimization mechanism at a time so the cause remains visible.
4. Verify required identifiers, tool calls, code changes or business fields.
5. Keep passthrough available when the reduced request does not pass.

Varion commercial evaluation

Varion keeps proprietary product implementation details private. Evaluate Token Optimisation on representative traffic and judge it by the measured commercial result. New verified users receive 100,000 processed input tokens and 50 local test runs.