The `reasoning_content` in the thinking mode must be passed back to the API.

[passwa11]

Description

Call deepseek api to report the following error
Bad Request: {"error":{"message":"The reasoning_content in the thinking mode must be passed back to the API.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

Plugins

No response

OpenCode version

No response

Steps to reproduce

No response

Screenshot and/or share link

No response

Operating System

No response

Terminal

No response

[github-actions]

This issue doesn't fully meet our contributing guidelines.

What needs to be fixed:

• Steps to reproduce is empty — please describe what you were doing when the error occurred (e.g., which model, what prompt style, whether extended thinking/reasoning mode was enabled)
• OpenCode version is missing — please provide the version you are running
• Operating System is missing — please include your OS and terminal info

Please edit this issue to address the above within 2 hours, or it will be automatically closed.

---

This issue might be a duplicate of existing issues. Please check:

• Missing reasoning_content field in the assistant message for DeepSeek API implementation #17523: Missing reasoning_content field in the assistant message for DeepSeek API implementation (open, same root cause)
• DeepSeek V3.2 : reasoning_content not cleared from message history on new turns, causing excess token usage and violating API spec #5577: DeepSeek V3.2 — reasoning_content not cleared from message history on new turns (closed, related)

If you believe this was flagged incorrectly, please let a maintainer know.

[Small-tailqwq]

存在同样的问题，使用deepseek-v4-pro模型时，使用/review触发子线程后，子线程任务结束时主线程报错.
The same issue exists. When using the deepseek-v4-pro model, after triggering a sub-thread with /review, the main thread reports an error when the sub-thread task ends.

The `reasoning_content` in the thinking mode must be passed back to the API.

情况如图所示 The situation is shown in the image.

下面是我的配置文件:
Below is my configuration file:：

{
  "npm": "@ai-sdk/openai-compatible",
  "options": {
    "baseURL": "https://api.deepseek.com/v1",
    "apiKey": "sk-**********************************"
  },
  "models": {
    "deepseek-v4-pro": {
      "name": "DeepSeek V4-Pro",
      "limit": {
        "context": 1048576,
        "output": 262144
      },
      "options": {
        "reasoningEffort": "max",
        "thinking": {
          "type": "enabled"
        }
      }
    },
    "deepseek-v4-flash": {
      "name": "deepseek-v4-flash"
    }
  }
}

此问题在多次测试过程中必定触发，不清楚是API提供商问题还是opencode处适配问题，用作信息补充。
This issue consistently occurs across multiple tests. It is unclear whether it is a problem with the API provider or an adaptation issue with opencode. This is provided as supplementary information.

---

补充说明（来自deepseek web 专家模式）
当你开启 thinking: { type: "enabled" } 后，API 返回的响应中会多一个 reasoning_content 字段（模型的思考过程）。为了保持多轮对话中思考的连续性，DeepSeek API 要求在下一轮请求中，必须把上一轮 assistant 消息里的 reasoning_content 原样传回。

如果在子线程 /review 结束后，主线程拼接历史消息时，丢失或没有正确传递这个 reasoning_content，API 就会报错：

The `reasoning_content` in the thinking mode must be passed back to the API.

解法1：
如果 opencode 支持自定义消息处理，需要检查它是否保存并在下次请求时携带了 reasoning_content。格式类似：

{
  "role": "assistant",
  "content": "模型的最终回复",
  "reasoning_content": "模型的思考过程（上一次API返回的这个字段）"
}

由于你提到不确定是 API 提供商还是 opencode 的适配问题，这很可能是 opencode 没有处理 thinking 模型的特殊消息结构。你可以向 opencode 提交 issue，附上你这段反馈，说明需要支持 DeepSeek 的 thinking 模式。

解法2：关闭 thinking 模式
如果不需要思维链，可以在配置中去掉 thinking 和 reasoningEffort：

"deepseek-v4-pro": {
  "name": "DeepSeek V4-Pro",
  "limit": {
    "context": 1048576,
    "output": 262144
  },
  "options": {
    // 删除或注释掉这两项
    // "reasoningEffort": "max",
    // "thinking": { "type": "enabled" }
  }
}

---

其他解决方案：来自#24097 (comment)
使用anthropic的接口格式来替换openai的接口格式：

{
  "npm": "@ai-sdk/anthropic",
  "name": "DeepSeek",
  "options": {
    "baseURL": "https://api.deepseek.com/anthropic",
    "apiKey": "sk-*****************"
  },
  "models": {
    "deepseek-v4-pro": {
      "name": "DeepSeek-V4-Pro",
      "limit": {
        "context": 1048576,
        "output": 262144
      },
      "options": {
        "thinking": {
          "type": "enabled",
          "budgetTokens": 8192
        }
      }
    },
    "deepseek-v4-flash": {
      "name": "DeepSeek-V4-Flash",
      "limit": {
        "context": 1048576,
        "output": 262144
      },
      "options": {
        "thinking": {
          "type": "enabled",
          "budgetTokens": 8192
        }
      }
    }
  }
}

此方案需要移除 reasoningEffort 参数，同时配合 budgetTokens 才能够正常使用。此方法只能作为临时解决方案，最终解决需要 opencode 添加对 deepseek 新模型的支持。
此方法经过测试会引出一个新的故障：

Bad Request: {"error":{"message":"The `content[].thinking` in the thinking mode must be passed back to the API.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

[devcxl]

存在同样的问题 我现在使用这个配置在会话中必定触发

"models": {
    "deepseek-v4-pro": {
      "name": "DeepSeek V4-Pro",
      "limit": {
        "context": 1048576,
        "output": 262144
      },
      "options": {
        "reasoningEffort": "max",
        "thinking": {
          "type": "enabled"
        }
      }
    },
    "deepseek-v4-flash": {
      "name": "deepseek-v4-flash"
    }
  }

[AzureHalcyon]

环境：windows11 25H2
版本：OpenCode Desktop v1.4.6
修改~/.config/opencode/opencode.json：

  "provider": {
    "deepseek": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "DeepSeek",
      "options": {
        "baseURL": "https://api.deepseek.com",
        "apiKey": "<DeepSeek API Key>"
      },
      "models": {
        "deepseek-v4-pro": {
          "name": "DeepSeek-V4-Pro",
          "limit": {
            "context": 1048576,
            "output": 262144
          },
          "options": {
            "reasoningEffort": "max",
            "thinking": {
              "type": "enabled"
            }
          }
        }
      }
    }
  }

问题：选择DeepSeek-V4-Pro进行多轮对话，几乎一定出现The reasoning_content in the thinking mode must be passed back to the API.报错。

[zhaojun8514-cmyk]

same issue

[github-actions]

This issue has been automatically closed because it was not updated to meet our contributing guidelines within the 2-hour window.

Feel free to open a new issue that follows our issue templates.