model-usage

You can get a per-model cost breakdown from local CodexBar logs by running the model_usage script with parameters like --provider codex --mode current to summarize usage data into readable text or JSON output.

Analyzing historical model usage costs for Codex and Claude is best handled by processing local cost logs with a script that supports full model breakdowns, returning a payload that includes model names and historical costs.

Yes, you can view the cost breakdown for just your current model by using the current mode, which specifically targets the active model's data instead of generating a full breakdown for all models.

The local cost summarizer works with missing data by gracefully handling gaps in the logs, validating inputs during processing to ensure the script returns available usage data without failing.

To export model usage summaries as a JSON payload for budget planning, run the local log summarizer to return a structured JSON object containing the model names and associated costs.