[Bug] API Key Misconfiguration - Custom Model Usage Consumes AIHubMix Tokens Instead of Personal Keys

[varcoh]

📦 部署方式

Other

📌 软件版本

V2.15.8

💻 系统环境

Ubuntu

📌 系统版本

24

🌐 浏览器

Edge

📌 浏览器版本

133.0.3065.82

🐛 问题描述

I am using an API Key obtained from AIHubMix and have set the BASE_URL in my .env file as follows:

BASE_URL=https://aihubmix.com

I have also added specific models in the custom settings, which work correctly. I can verify that token consumption details appear on AIHubMix.

However, I have separately applied for a DeepSeek API Key and configured it in my .env file:

DEEPSEEK_API_KEY=sk-XXXXX
DEEPSEEK_URL=https://api.deepseek.com

I then added the following custom model setting:

+deepseek-chat=DeepSeek

But when I use DeepSeek, it still consumes tokens from AIHubMix instead of my personal DeepSeek API Key.

Similarly, I applied for a Grok API Key and configured it as follows:

GROK_API_KEY=xai-XXXXX
GROK_URL=https://api.x.ai/v1

However, when using Grok, it also consumes AIHubMix tokens instead of my personal Grok API Key.

Where could the issue be? Is there an additional configuration required to ensure that DeepSeek and Grok use my personal API keys instead of AIHubMix?

Looking forward to your guidance. Thanks!

📷 复现步骤

No response

🚦 期望结果

No response

📝 补充信息

No response