How can I use the GPT-4 Vision API 'detail' parameter and retrieve token cost information?

Below is a summary of how things currently work according to our documentation:

---

Using the detail Parameter

• The API accepts a detail option for each image URL block. You can specify either "low" or "high".

  • Low detail: The pricing is fixed at 85 tokens regardless of the image’s dimensions.
  • High detail: The image is first scaled to fit within a 2048×2048 square, then further reduced (with the shortest side becoming 768px) so that the cost is calculated based on how many 512px square tiles the image contains. Each tile costs 170 tokens, plus an additional 85 tokens are added to the final cost.

• This parameter helps you manage your usage and expense—you control the trade-off between image fidelity and token usage.

---

Retrieving Token Cost Information

• The API returns token counts as part of its response. This allows you to see exactly how much of your token quota was used for a given request.
• For cost estimations, our documentation recommends using the interactive tokenizer tool, which can be helpful for both text and image input cases.

---

Quick Notes on Early Adoption

• Our documentation reflects the current pricing model and the accepted parameters. That said, some aspects might be experimental or subject to revision as we continue to iterate.
• If you’re integrating this functionality, it’s a good idea to monitor the official documentation for any updates.

In summary, the detail parameter should work as described, and your API responses include token cost details so you can track usage efficiently. If you run into issues with sending the parameter or retrieving token counts, double-check your request structure against the latest documentation and ensure your integration conforms to the expected JSON schema.

Let me know if you need further details on integrating these features.