agno 2.0.0rc2__py3-none-any.whl → 2.0.2__py3-none-any.whl

agno/models/base.py

@@ -21,7 +21,7 @@ from uuid import uuid4
 from pydantic import BaseModel
 
 from agno.exceptions import AgentRunException
-from agno.media import Audio, AudioArtifact, AudioResponse, Image, ImageArtifact, Video, VideoArtifact
+from agno.media import Audio, Image, Video
 from agno.models.message import Citations, Message
 from agno.models.metrics import Metrics
 from agno.models.response import ModelResponse, ModelResponseEvent, ToolExecution
@@ -43,9 +43,9 @@ class MessageData:
     response_citations: Optional[Citations] = None
     response_tool_calls: List[Dict[str, Any]] = field(default_factory=list)
 
-    response_audio: Optional[AudioResponse] = None
-    response_image: Optional[ImageArtifact] = None
-    response_video: Optional[VideoArtifact] = None
+    response_audio: Optional[Audio] = None
+    response_image: Optional[Image] = None
+    response_video: Optional[Video] = None
 
     # Data from the provider that we might need on subsequent messages
     response_provider_data: Optional[Dict[str, Any]] = None
@@ -502,9 +502,7 @@ class Model(ABC):
         if assistant_message.citations is not None:
             model_response.citations = assistant_message.citations
         if assistant_message.audio_output is not None:
-            if isinstance(assistant_message.audio_output, AudioArtifact):
-                model_response.audios = [assistant_message.audio_output]
-            elif isinstance(assistant_message.audio_output, AudioResponse):
+            if isinstance(assistant_message.audio_output, Audio):
                 model_response.audio = assistant_message.audio_output
         if assistant_message.image_output is not None:
             model_response.images = [assistant_message.image_output]
@@ -557,9 +555,7 @@ class Model(ABC):
         if assistant_message.citations is not None:
             model_response.citations = assistant_message.citations
         if assistant_message.audio_output is not None:
-            if isinstance(assistant_message.audio_output, AudioArtifact):
-                model_response.audios = [assistant_message.audio_output]
-            elif isinstance(assistant_message.audio_output, AudioResponse):
+            if isinstance(assistant_message.audio_output, Audio):
                 model_response.audio = assistant_message.audio_output
         if assistant_message.image_output is not None:
             model_response.images = [assistant_message.image_output]
@@ -993,13 +989,13 @@ class Model(ABC):
             stream_data.response_tool_calls.extend(model_response_delta.tool_calls)
             should_yield = True
 
-        if model_response_delta.audio is not None and isinstance(model_response_delta.audio, AudioResponse):
+        if model_response_delta.audio is not None and isinstance(model_response_delta.audio, Audio):
             if stream_data.response_audio is None:
-                stream_data.response_audio = AudioResponse(id=str(uuid4()), content="", transcript="")
+                stream_data.response_audio = Audio(id=str(uuid4()), content="", transcript="")
 
             from typing import cast
 
-            audio_response = cast(AudioResponse, model_response_delta.audio)
+            audio_response = cast(Audio, model_response_delta.audio)
 
             # Update the stream data with audio information
             if audio_response.id is not None:
@@ -1104,41 +1100,10 @@ class Model(ABC):
         audios = None
 
         if success and function_execution_result:
-            # Convert ImageArtifacts to Images for message compatibility
-            if function_execution_result.images:
-                from agno.media import Image
-
-                images = []
-                for img_artifact in function_execution_result.images:
-                    if img_artifact.url:
-                        images.append(Image(url=img_artifact.url))
-                    elif img_artifact.content:
-                        images.append(Image(content=img_artifact.content))
-
-            # Convert VideoArtifacts to Videos for message compatibility
-            if function_execution_result.videos:
-                from agno.media import Video
-
-                videos = []
-                for vid_artifact in function_execution_result.videos:
-                    if vid_artifact.url:
-                        videos.append(Video(url=vid_artifact.url))
-                    elif vid_artifact.content:
-                        videos.append(Video(content=vid_artifact.content))
-
-            # Convert AudioArtifacts to Audio for message compatibility
-            if function_execution_result.audios:
-                from agno.media import Audio
-
-                audios = []
-                for aud_artifact in function_execution_result.audios:
-                    if aud_artifact.url:
-                        audios.append(Audio(url=aud_artifact.url))
-                    elif aud_artifact.base64_audio:
-                        import base64
-
-                        audio_bytes = base64.b64decode(aud_artifact.base64_audio)
-                        audios.append(Audio(content=audio_bytes))
+            # With unified classes, no conversion needed - use directly
+            images = function_execution_result.images
+            videos = function_execution_result.videos
+            audios = function_execution_result.audios
 
         return Message(
             role=self.tool_message_role,

agno/models/google/gemini.py

@@ -10,7 +10,7 @@ from uuid import uuid4
 from pydantic import BaseModel
 
 from agno.exceptions import ModelProviderError
-from agno.media import Audio, File, ImageArtifact, Video
+from agno.media import Audio, File, Image, Video
 from agno.models.base import Model
 from agno.models.message import Citations, Message, UrlCitation
 from agno.models.metrics import Metrics
@@ -559,9 +559,14 @@ class Gemini(Model):
             return Part.from_bytes(mime_type=mime_type, data=audio.content)
 
         # Case 2: Audio is an url
-        elif audio.url is not None and audio.audio_url_content is not None:
-            mime_type = f"audio/{audio.format}" if audio.format else "audio/mp3"
-            return Part.from_bytes(mime_type=mime_type, data=audio.audio_url_content)
+        elif audio.url is not None:
+            audio_bytes = audio.get_content_bytes()  # type: ignore
+            if audio_bytes is not None:
+                mime_type = f"audio/{audio.format}" if audio.format else "audio/mp3"
+                return Part.from_bytes(mime_type=mime_type, data=audio_bytes)
+            else:
+                log_warning(f"Failed to download audio from {audio}")
+                return None
 
         # Case 3: Audio is a local file path
         elif audio.filepath is not None:
@@ -815,9 +820,7 @@ class Gemini(Model):
                     if model_response.images is None:
                         model_response.images = []
                     model_response.images.append(
-                        ImageArtifact(
-                            id=str(uuid4()), content=part.inline_data.data, mime_type=part.inline_data.mime_type
-                        )
+                        Image(id=str(uuid4()), content=part.inline_data.data, mime_type=part.inline_data.mime_type)
                     )
 
                 # Extract function call if present
@@ -929,9 +932,7 @@ class Gemini(Model):
                         if model_response.images is None:
                             model_response.images = []
                         model_response.images.append(
-                            ImageArtifact(
-                                id=str(uuid4()), content=part.inline_data.data, mime_type=part.inline_data.mime_type
-                            )
+                            Image(id=str(uuid4()), content=part.inline_data.data, mime_type=part.inline_data.mime_type)
                         )
 
                     # Extract function call if present

agno/models/message.py

@@ -4,7 +4,7 @@ from typing import Any, Dict, List, Optional, Sequence, Union
 
 from pydantic import BaseModel, ConfigDict, Field
 
-from agno.media import Audio, AudioArtifact, AudioResponse, File, Image, ImageArtifact, Video, VideoArtifact
+from agno.media import Audio, File, Image, Video
 from agno.models.metrics import Metrics
 from agno.utils.log import log_debug, log_error, log_info, log_warning
 
@@ -71,9 +71,9 @@ class Message(BaseModel):
     files: Optional[Sequence[File]] = None
 
     # Output from the models
-    audio_output: Optional[Union[AudioResponse, AudioArtifact]] = None
-    image_output: Optional[ImageArtifact] = None
-    video_output: Optional[VideoArtifact] = None
+    audio_output: Optional[Audio] = None
+    image_output: Optional[Image] = None
+    video_output: Optional[Video] = None
 
     # The thinking content from the model
     redacted_reasoning_content: Optional[str] = None

agno/models/ollama/chat.py

@@ -149,7 +149,7 @@ class Ollama(Model):
                 message_images = []
                 for image in message.images:
                     if image.url is not None:
-                        message_images.append(image.image_url_content)
+                        message_images.append(image.get_content_bytes())
                     if image.filepath is not None:
                         message_images.append(image.filepath)  # type: ignore
                     if image.content is not None and isinstance(image.content, bytes):

agno/models/openai/chat.py

@@ -2,12 +2,13 @@ from collections.abc import AsyncIterator
 from dataclasses import dataclass
 from os import getenv
 from typing import Any, Dict, Iterator, List, Literal, Optional, Type, Union
+from uuid import uuid4
 
 import httpx
 from pydantic import BaseModel
 
 from agno.exceptions import ModelProviderError
-from agno.media import AudioResponse
+from agno.media import Audio
 from agno.models.base import Model
 from agno.models.message import Message
 from agno.models.metrics import Metrics
@@ -729,14 +730,14 @@ class OpenAIChat(Model):
             # If the audio output modality is requested, we can extract an audio response
             try:
                 if isinstance(response_message.audio, dict):
-                    model_response.audio = AudioResponse(
+                    model_response.audio = Audio(
                         id=response_message.audio.get("id"),
                         content=response_message.audio.get("data"),
                         expires_at=response_message.audio.get("expires_at"),
                         transcript=response_message.audio.get("transcript"),
                     )
                 else:
-                    model_response.audio = AudioResponse(
+                    model_response.audio = Audio(
                         id=response_message.audio.id,
                         content=response_message.audio.data,
                         expires_at=response_message.audio.expires_at,
@@ -783,21 +784,39 @@ class OpenAIChat(Model):
                 # Add audio if present
                 if hasattr(choice_delta, "audio") and choice_delta.audio is not None:
                     try:
+                        audio_data = None
+                        audio_id = None
+                        audio_expires_at = None
+                        audio_transcript = None
+
                         if isinstance(choice_delta.audio, dict):
-                            model_response.audio = AudioResponse(
-                                id=choice_delta.audio.get("id"),
-                                content=choice_delta.audio.get("data"),
-                                expires_at=choice_delta.audio.get("expires_at"),
-                                transcript=choice_delta.audio.get("transcript"),
+                            audio_data = choice_delta.audio.get("data")
+                            audio_id = choice_delta.audio.get("id")
+                            audio_expires_at = choice_delta.audio.get("expires_at")
+                            audio_transcript = choice_delta.audio.get("transcript")
+                        else:
+                            audio_data = choice_delta.audio.data
+                            audio_id = choice_delta.audio.id
+                            audio_expires_at = choice_delta.audio.expires_at
+                            audio_transcript = choice_delta.audio.transcript
+
+                        # Only create Audio object if there's actual content
+                        if audio_data is not None:
+                            model_response.audio = Audio(
+                                id=audio_id,
+                                content=audio_data,
+                                expires_at=audio_expires_at,
+                                transcript=audio_transcript,
                                 sample_rate=24000,
                                 mime_type="pcm16",
                             )
-                        else:
-                            model_response.audio = AudioResponse(
-                                id=choice_delta.audio.id,
-                                content=choice_delta.audio.data,
-                                expires_at=choice_delta.audio.expires_at,
-                                transcript=choice_delta.audio.transcript,
+                        # If no content but there's transcript/metadata, create minimal Audio object
+                        elif audio_transcript is not None or audio_id is not None:
+                            model_response.audio = Audio(
+                                id=audio_id or str(uuid4()),
+                                content=b"",
+                                expires_at=audio_expires_at,
+                                transcript=audio_transcript,
                                 sample_rate=24000,
                                 mime_type="pcm16",
                             )

agno/models/response.py

@@ -3,7 +3,7 @@ from enum import Enum
 from time import time
 from typing import Any, Dict, List, Optional
 
-from agno.media import AudioArtifact, AudioResponse, ImageArtifact, VideoArtifact
+from agno.media import Audio, Image, Video
 from agno.models.message import Citations
 from agno.models.metrics import Metrics
 from agno.tools.function import UserInputField
@@ -87,12 +87,12 @@ class ModelResponse:
 
     content: Optional[Any] = None
     parsed: Optional[Any] = None
-    audio: Optional[AudioResponse] = None
+    audio: Optional[Audio] = None
 
     # Unified media fields for LLM-generated and tool-generated media artifacts
-    images: Optional[List[ImageArtifact]] = None
-    videos: Optional[List[VideoArtifact]] = None
-    audios: Optional[List[AudioArtifact]] = None
+    images: Optional[List[Image]] = None
+    videos: Optional[List[Video]] = None
+    audios: Optional[List[Audio]] = None
 
     # Model tool calls
     tool_calls: List[Dict[str, Any]] = field(default_factory=list)

agno/os/app.py

@@ -36,12 +36,11 @@ from agno.os.routers.session import get_session_router
 from agno.os.settings import AgnoAPISettings
 from agno.os.utils import generate_id
 from agno.team.team import Team
-from agno.tools.mcp import MCPTools, MultiMCPTools
 from agno.workflow.workflow import Workflow
 
 
 @asynccontextmanager
-async def mcp_lifespan(app, mcp_tools: List[Union[MCPTools, MultiMCPTools]]):
+async def mcp_lifespan(app, mcp_tools):
     """Manage MCP connection lifecycle inside a FastAPI app"""
     # Startup logic: connect to all contextual MCP servers
     for tool in mcp_tools:
@@ -103,14 +102,16 @@ class AgentOS:
         self.lifespan = lifespan
 
         # List of all MCP tools used inside the AgentOS
-        self.mcp_tools: List[Union[MCPTools, MultiMCPTools]] = []
+        self.mcp_tools = []
 
         if self.agents:
             for agent in self.agents:
                 # Track all MCP tools to later handle their connection
                 if agent.tools:
                     for tool in agent.tools:
-                        if isinstance(tool, MCPTools) or isinstance(tool, MultiMCPTools):
+                        # Checking if the tool is a MCPTools or MultiMCPTools instance
+                        type_name = type(tool).__name__
+                        if type_name in ("MCPTools", "MultiMCPTools"):
                             self.mcp_tools.append(tool)
 
                 agent.initialize_agent()
@@ -123,7 +124,9 @@ class AgentOS:
                 # Track all MCP tools to later handle their connection
                 if team.tools:
                     for tool in team.tools:
-                        if isinstance(tool, MCPTools) or isinstance(tool, MultiMCPTools):
+                        # Checking if the tool is a MCPTools or MultiMCPTools instance
+                        type_name = type(tool).__name__
+                        if type_name in ("MCPTools", "MultiMCPTools"):
                             self.mcp_tools.append(tool)
 
                 team.initialize_team()

agno/run/agent.py

@@ -5,7 +5,7 @@ from typing import Any, Dict, List, Optional, Sequence, Union
 
 from pydantic import BaseModel
 
-from agno.media import AudioArtifact, AudioResponse, File, ImageArtifact, VideoArtifact
+from agno.media import Audio, File, Image, Video
 from agno.models.message import Citations, Message
 from agno.models.metrics import Metrics
 from agno.models.response import ToolExecution
@@ -97,8 +97,8 @@ class RunContentEvent(BaseAgentRunEvent):
     content_type: str = "str"
     reasoning_content: Optional[str] = None
     citations: Optional[Citations] = None
-    response_audio: Optional[AudioResponse] = None  # Model audio response
-    image: Optional[ImageArtifact] = None  # Image attached to the response
+    response_audio: Optional[Audio] = None  # Model audio response
+    image: Optional[Image] = None  # Image attached to the response
     references: Optional[List[MessageReferences]] = None
     additional_input: Optional[List[Message]] = None
     reasoning_steps: Optional[List[ReasoningStep]] = None
@@ -119,10 +119,10 @@ class RunCompletedEvent(BaseAgentRunEvent):
     content_type: str = "str"
     reasoning_content: Optional[str] = None
     citations: Optional[Citations] = None
-    images: Optional[List[ImageArtifact]] = None  # Images attached to the response
-    videos: Optional[List[VideoArtifact]] = None  # Videos attached to the response
-    audio: Optional[List[AudioArtifact]] = None  # Audio attached to the response
-    response_audio: Optional[AudioResponse] = None  # Model audio response
+    images: Optional[List[Image]] = None  # Images attached to the response
+    videos: Optional[List[Video]] = None  # Videos attached to the response
+    audio: Optional[List[Audio]] = None  # Audio attached to the response
+    response_audio: Optional[Audio] = None  # Model audio response
     references: Optional[List[MessageReferences]] = None
     additional_input: Optional[List[Message]] = None
     reasoning_steps: Optional[List[ReasoningStep]] = None
@@ -203,9 +203,9 @@ class ToolCallCompletedEvent(BaseAgentRunEvent):
     event: str = RunEvent.tool_call_completed.value
     tool: Optional[ToolExecution] = None
     content: Optional[Any] = None
-    images: Optional[List[ImageArtifact]] = None  # Images produced by the tool call
-    videos: Optional[List[VideoArtifact]] = None  # Videos produced by the tool call
-    audio: Optional[List[AudioArtifact]] = None  # Audio produced by the tool call
+    images: Optional[List[Image]] = None  # Images produced by the tool call
+    videos: Optional[List[Video]] = None  # Videos produced by the tool call
+    audio: Optional[List[Audio]] = None  # Audio produced by the tool call
 
 
 @dataclass
@@ -306,9 +306,9 @@ class RunInput:
     """
 
     input_content: Optional[Union[str, List, Dict, Message, BaseModel, List[Message]]] = None
-    images: Optional[Sequence[ImageArtifact]] = None
-    videos: Optional[Sequence[VideoArtifact]] = None
-    audios: Optional[Sequence[AudioArtifact]] = None
+    images: Optional[Sequence[Image]] = None
+    videos: Optional[Sequence[Video]] = None
+    audios: Optional[Sequence[Audio]] = None
     files: Optional[Sequence[File]] = None
 
     def to_dict(self) -> Dict[str, Any]:
@@ -345,15 +345,15 @@ class RunInput:
         """Create RunInput from dictionary"""
         images = None
         if data.get("images"):
-            images = [ImageArtifact.model_validate(img_data) for img_data in data["images"]]
+            images = [Image.model_validate(img_data) for img_data in data["images"]]
 
         videos = None
         if data.get("videos"):
-            videos = [VideoArtifact.model_validate(vid_data) for vid_data in data["videos"]]
+            videos = [Video.model_validate(vid_data) for vid_data in data["videos"]]
 
         audios = None
         if data.get("audios"):
-            audios = [AudioArtifact.model_validate(aud_data) for aud_data in data["audios"]]
+            audios = [Audio.model_validate(aud_data) for aud_data in data["audios"]]
 
         files = None
         if data.get("files"):
@@ -389,10 +389,10 @@ class RunOutput:
 
     tools: Optional[List[ToolExecution]] = None
 
-    images: Optional[List[ImageArtifact]] = None  # Images attached to the response
-    videos: Optional[List[VideoArtifact]] = None  # Videos attached to the response
-    audio: Optional[List[AudioArtifact]] = None  # Audio attached to the response
-    response_audio: Optional[AudioResponse] = None  # Model audio response
+    images: Optional[List[Image]] = None  # Images attached to the response
+    videos: Optional[List[Video]] = None  # Videos attached to the response
+    audio: Optional[List[Audio]] = None  # Audio attached to the response
+    response_audio: Optional[Audio] = None  # Model audio response
 
     # Input media and messages from user
     input: Optional[RunInput] = None
@@ -487,7 +487,7 @@ class RunOutput:
         if self.images is not None:
             _dict["images"] = []
             for img in self.images:
-                if isinstance(img, ImageArtifact):
+                if isinstance(img, Image):
                     _dict["images"].append(img.to_dict())
                 else:
                     _dict["images"].append(img)
@@ -495,7 +495,7 @@ class RunOutput:
         if self.videos is not None:
             _dict["videos"] = []
             for vid in self.videos:
-                if isinstance(vid, VideoArtifact):
+                if isinstance(vid, Video):
                     _dict["videos"].append(vid.to_dict())
                 else:
                     _dict["videos"].append(vid)
@@ -503,13 +503,13 @@ class RunOutput:
         if self.audio is not None:
             _dict["audio"] = []
             for aud in self.audio:
-                if isinstance(aud, AudioArtifact):
+                if isinstance(aud, Audio):
                     _dict["audio"].append(aud.to_dict())
                 else:
                     _dict["audio"].append(aud)
 
         if self.response_audio is not None:
-            if isinstance(self.response_audio, AudioResponse):
+            if isinstance(self.response_audio, Audio):
                 _dict["response_audio"] = self.response_audio.to_dict()
             else:
                 _dict["response_audio"] = self.response_audio
@@ -565,16 +565,16 @@ class RunOutput:
         tools = [ToolExecution.from_dict(tool) for tool in tools] if tools else None
 
         images = data.pop("images", [])
-        images = [ImageArtifact.model_validate(image) for image in images] if images else None
+        images = [Image.model_validate(image) for image in images] if images else None
 
         videos = data.pop("videos", [])
-        videos = [VideoArtifact.model_validate(video) for video in videos] if videos else None
+        videos = [Video.model_validate(video) for video in videos] if videos else None
 
         audio = data.pop("audio", [])
-        audio = [AudioArtifact.model_validate(audio) for audio in audio] if audio else None
+        audio = [Audio.model_validate(audio) for audio in audio] if audio else None
 
         response_audio = data.pop("response_audio", None)
-        response_audio = AudioResponse.model_validate(response_audio) if response_audio else None
+        response_audio = Audio.model_validate(response_audio) if response_audio else None
 
         input_data = data.pop("input", None)
         input_obj = None

agno/run/base.py

@@ -4,7 +4,7 @@ from typing import Any, Dict
 
 from pydantic import BaseModel
 
-from agno.media import AudioArtifact, AudioResponse, ImageArtifact, VideoArtifact
+from agno.media import Audio, Image, Video
 from agno.models.message import Citations, Message, MessageReferences
 from agno.models.metrics import Metrics
 from agno.models.response import ToolExecution
@@ -60,21 +60,15 @@ class BaseRunOutputEvent:
         if hasattr(self, "images") and self.images is not None:
             _dict["images"] = []
             for img in self.images:
-                if isinstance(img, ImageArtifact):
+                if isinstance(img, Image):
                     _dict["images"].append(img.to_dict())
                 else:
                     _dict["images"].append(img)
 
-        if hasattr(self, "image") and self.image is not None:
-            if isinstance(self.image, ImageArtifact):
-                _dict["image"] = self.image.to_dict()
-            else:
-                _dict["image"] = self.image
-
         if hasattr(self, "videos") and self.videos is not None:
             _dict["videos"] = []
             for vid in self.videos:
-                if isinstance(vid, VideoArtifact):
+                if isinstance(vid, Video):
                     _dict["videos"].append(vid.to_dict())
                 else:
                     _dict["videos"].append(vid)
@@ -82,13 +76,13 @@ class BaseRunOutputEvent:
         if hasattr(self, "audio") and self.audio is not None:
             _dict["audio"] = []
             for aud in self.audio:
-                if isinstance(aud, AudioArtifact):
+                if isinstance(aud, Audio):
                     _dict["audio"].append(aud.to_dict())
                 else:
                     _dict["audio"].append(aud)
 
         if hasattr(self, "response_audio") and self.response_audio is not None:
-            if isinstance(self.response_audio, AudioResponse):
+            if isinstance(self.response_audio, Audio):
                 _dict["response_audio"] = self.response_audio.to_dict()
             else:
                 _dict["response_audio"] = self.response_audio
@@ -140,23 +134,19 @@ class BaseRunOutputEvent:
 
         images = data.pop("images", None)
         if images:
-            data["images"] = [ImageArtifact.model_validate(image) for image in images]
-
-        image = data.pop("image", None)
-        if image:
-            data["image"] = ImageArtifact.model_validate(image)
+            data["images"] = [Image.model_validate(image) for image in images]
 
         videos = data.pop("videos", None)
         if videos:
-            data["videos"] = [VideoArtifact.model_validate(video) for video in videos]
+            data["videos"] = [Video.model_validate(video) for video in videos]
 
         audio = data.pop("audio", None)
         if audio:
-            data["audio"] = [AudioArtifact.model_validate(audio) for audio in audio]
+            data["audio"] = [Audio.model_validate(audio) for audio in audio]
 
         response_audio = data.pop("response_audio", None)
         if response_audio:
-            data["response_audio"] = AudioResponse.model_validate(response_audio)
+            data["response_audio"] = Audio.model_validate(response_audio)
 
         additional_input = data.pop("additional_input", None)
         if additional_input is not None: