Use plan.md to compare AI model speed and cost
A plan.md file lists five varied tasks that match normal daily work. Different AI models, including newer ones such as and existing choices, can run the same tasks for comparison. The method tracks not only whether the result is good, but also how long the work takes and how much it costs.
After the work is done, checks whether the output is actually correct. This is an intermediate benchmark workflow for comparing model , time, and cost in real .
Key points
- Write five realistic work tasks in a plan.md file.
- Run the same tasks with different AI models.
- Record result quality, time, and cost for each run.
- Use to check correctness.
- Use the records to decide which model fits a specific workflow.