NexAIX-官方博客
Adding an OpenAI-Compatible Model in Dify: How to Fill the base_url, Whether to Include /v1, and How to Troubleshoot Save Failures
In Dify's OpenAI-API-compatible provider, the LLM API endpoint URL must be written as "protocol + domain + /v1" with no trailing slash; this article explains how to fill each field and how to troubleshoot save failures such as 404s from missing /v1, /v1/v1 on non-LLM models, mismatched model names, and misused localhost in Docker.
DeepSeek Chat Alias Retirement: OpenAI-Compatible API Model Migration Path and Regression Checklist
Learn how to migrate OpenAI-compatible API models after DeepSeek retires old aliases, with a practical regression checklist.
