Skip to content

ImageGenerationTool

autogen.beta.tools.builtin.image_generation.ImageGenerationTool #

ImageGenerationTool(*, quality=None, size=None, background=None, output_format=None, output_compression=None, partial_images=None)

Bases: Tool

Builtin image generation tool for the OpenAI Responses API.

Instructs the model to generate images inline during the conversation. Only supported with OpenAIResponsesConfig — raises UnsupportedToolError when used with OpenAIConfig (Chat Completions API).

Generated images are returned as list[BinaryResult] via reply.files.

PARAMETER DESCRIPTION
quality

Image quality — "low", "medium", "high", or "auto".

TYPE: Literal['low', 'medium', 'high', 'auto'] | Variable | None DEFAULT: None

size

Image dimensions, e.g. "1024x1024", "1536x1024", "auto".

TYPE: str | Variable | None DEFAULT: None

background

Background type — "transparent", "opaque", or "auto".

TYPE: Literal['transparent', 'opaque', 'auto'] | Variable | None DEFAULT: None

output_format

Output format — "png", "jpeg", or "webp".

TYPE: Literal['png', 'jpeg', 'webp'] | Variable | None DEFAULT: None

output_compression

Compression level 0–100 (webp/jpeg only).

TYPE: int | Variable | None DEFAULT: None

partial_images

Number of partial images to stream (1–3).

TYPE: int | Variable | None DEFAULT: None

Source code in autogen/beta/tools/builtin/image_generation.py 
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
def __init__(
    self,
    *,
    quality: Literal["low", "medium", "high", "auto"] | Variable | None = None,
    size: str | Variable | None = None,
    background: Literal["transparent", "opaque", "auto"] | Variable | None = None,
    output_format: Literal["png", "jpeg", "webp"] | Variable | None = None,
    output_compression: int | Variable | None = None,
    partial_images: int | Variable | None = None,
) -> None:
    self._params: dict[str, object] = {}
    if quality is not None:
        self._params["quality"] = quality
    if size is not None:
        self._params["size"] = size
    if background is not None:
        self._params["background"] = background
    if output_format is not None:
        self._params["output_format"] = output_format
    if output_compression is not None:
        self._params["output_compression"] = output_compression
    if partial_images is not None:
        self._params["partial_images"] = partial_images

    self.name = IMAGE_GENERATION_TOOL_NAME

name instance-attribute #

name = IMAGE_GENERATION_TOOL_NAME

schemas async #

schemas(context)
Source code in autogen/beta/tools/builtin/image_generation.py 
83
84
85
async def schemas(self, context: "Context") -> list[ImageGenerationToolSchema]:
    resolved = {k: resolve_variable(v, context, param_name=k) for k, v in self._params.items()}
    return [ImageGenerationToolSchema(**resolved)]

register #

register(stack, context, *, middleware=())
Source code in autogen/beta/tools/builtin/image_generation.py 
87
88
89
90
91
92
93
94
95
96
97
98
99
def register(
    self,
    stack: "ExitStack | AsyncExitStack",
    context: "Context",
    *,
    middleware: Iterable["BaseMiddleware"] = (),
) -> None:
    async def execute(event: "ToolCallEvent", context: "Context") -> None:
        pass

    stack.enter_context(
        context.stream.where(BuiltinToolCallEvent.name == IMAGE_GENERATION_TOOL_NAME).sub_scope(execute),
    )

set_provider #

set_provider(provider)
Source code in autogen/beta/tools/tool.py 
20
21
def set_provider(self, provider: Provider) -> None:
    pass