From a205114d7d92f2fee4de9c36fb88cd6fe860977f Mon Sep 17 00:00:00 2001
From: cscguochang-agent <cscguochang@gmail.com>
Date: Sun, 1 Feb 2026 03:22:58 +0800
Subject: [PATCH] fix(deepseek): add support for prompt_tokens_details in usage
 response

- Properly maps 'prompt_cache_hit_tokens' and 'prompt_cache_miss_tokens' to OpenAI-compatible 'prompt_tokens_details'.
- Fixes Issue #16009 for commercial billing accuracy in proxies.
- Ensures cross-provider consistency for cached token tracking.