Skip to content

接不支持视觉信息的LLM #8

Description

@a1443356159

相关逻辑在 agent/backends/planner.py:785。当视觉开启时,它最终生成类似 OpenAI 的多模态消息:

{
"type": "image_url",
"image_url": {
"url": "data:image/png;base64,..."
}
}

DeepSeek 当前这个 endpoint 只接受:

{"type": "text"}

所以返回了:

unknown variant image_url, expected text

原来的问题是:底层虽然有 enable_vision,但 .env 和 TUI 没有暴露这个配置。

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions