model-usage

You can aggregate per-model usage costs from CodexBar local logs by parsing local JSON files, extracting cost data from modelBreakdowns and modelsUsed, and calculating totals for budgeting and reporting.

Summarizing Codex and Claude model costs is achieved by reading local JSON cost logs, aggregating per-model expenses, and outputting either human-readable text or JSON with totals and latest-day costs.

Yes, you can generate a current-model summary or a full all-model breakdown from the aggregated cost data, allowing you to focus on specific model usage or view comprehensive reporting.

This approach supports both plain text and JSON output formats, fitting seamlessly into automated pipelines or displaying human-readable summaries directly on monitoring dashboards.

Parsing modelBreakdowns and modelsUsed is necessary because these fields within local JSON logs contain the granular per-model cost data required to calculate accurate totals and latest-day expenses.

Extracting latest-day costs and date stamps involves reading local JSON cost logs, aggregating the modelBreakdowns data, and formatting the results into a structured output with the corresponding date stamps.