Hello everyone!
I want to use the DeepSeek R1 (Reasoner) model within the native Make AI Agent module to build an autonomous agent that handles complex logic and reasoning.
I am encountering a persistent 400 Bad Request error when using deepseek-reasoner. The issue seems to be that the Make AI Agent module does not store or send back the reasoning_content field required by the DeepSeek R1 API for conversation history context.
What I have tried:
I switched the model to deepseek-chat (V3) in the same Agent module configuration, and it works perfectly. This confirms my connection and API key are correct.
I attempted to use deepseek-reasoner again, but it fails as soon as the conversation has a second turn or a tool call (likely due to the missing reasoning history).
Error messages or input/output bundles
400 Missing reasoning_content field in the assistant message at message index 2. For more information, please refer to Thinking Mode | DeepSeek API Docs
Screenshots (scenario flow, module settings, errors)
