Pricing and Cost Checks

Before testing long-context or agent workflows, check the current model price at:

https://lizh.ai/pricing

Basic cost habits

  • Start with a short prompt.
  • Set a small output limit.
  • Review usage logs after the first request.
  • Compare input tokens, output tokens, and cache-hit tokens separately.
  • Increase prompt size only after the small request works.

Why this matters

Coding agents and long-context prompts can send large inputs. A small cURL or SDK test tells you whether the configuration works before you spend tokens on a full workflow.

Explain what this function does in 5 bullets.

Then move to larger tasks such as:

  • repository summaries,
  • file-by-file review,
  • test generation,
  • error-log analysis,
  • refactoring plans.