#8448 You selected LLM error correction and used the Local LLM (OpenAI-compatible) channel, but did not fill in the key or api

This is because LLM Correction is enabled and the LLM correction channel is set to Local LLM (OpenAI-compatible), but the API URL (and key) for that channel have not been configured. The program therefore refuses to run.

Choose one of the following:

Option 1: Fill in the Local LLM configuration (if you really want to use a local model)

1. Start your local OpenAI-compatible service (Ollama / LM Studio / vLLM, etc.) and confirm the endpoint address, e.g. http://127.0.0.1:11434/v1.
2. Open Menu → Translation Settings → Compatible AI / Local LLM (i.e. the Local LLM channel settings).
3. Fill in the API URL (must include the /v1 suffix, and 0.0.0.0 must be written as 127.0.0.1); the key can be any non-empty string (some local services accept any value, but the field cannot be left blank).
4. Fill in the model name, save, then run the task again.

Option 2: Switch the correction channel to an online AI (simplest)

1. Open Menu → Tools/Options → Advanced Options → Speech Recognition (ASR) → LLM Provider for Proofreading.
2. Change it to DeepSeek (default) or another AI channel whose key you have already configured, and save.

Option 3: Do not use LLM correction

1. On the main interface, uncheck LLM Correction in the STT row.

Please refer to the related documentation:

• Local LLM / domestic AI integration
• Advanced Options