DeepSeek V4 "reasoning_content" error (400) during multi-turn chat

[subediparas5]

What happened?

When using the deepseek-v4-pro or deepseek-v4-flash models, the API returns a 400 error: The 'reasoning_content' in the thinking mode must be passed back to the API.

Steps to reproduce

1. Select a DeepSeek V4 model.
2. Provide a complex prompt (e.g., "Optimize this SQL query").
3. The model generates "thinking" or "reasoning" content.
4. On the subsequent message (or when the app tries to follow up with tool calls), the API throws a 400 error because the reasoning_content from the previous assistant message was not included in the request payload.

Expected behavior

No response

Database type

MySQL / MariaDB

TablePro version

No response

macOS version & chip

No response

Screenshots / Logs