Merge pull request #4447 from pipecat-ai/pk/realtime-async-tool-support-followup

fix: extend cancel_on_interruption=False regression fix to remaining realtime services

changelog/4447.fixed.md

@@ -0,0 +1 @@
+- Extended the `cancel_on_interruption=False` regression fix to `GrokRealtimeLLMService`, `AzureRealtimeLLMService`, and `UltravoxRealtimeLLMService`. Grok and Azure use the same approach as in #4441 (each service detects async-tool messages in the LLM context and routes the final result to its formal tool-result channel; Azure inherits transitively from `OpenAIRealtimeLLMService`). Ultravox needed a different approach because its API freezes the conversation between `client_tool_invocation` and the matching `client_tool_result` — for async-registered functions it now ships a placeholder `client_tool_result` immediately when the function is invoked (to unfreeze the conversation), then injects the real result as user-side text once the tool finishes. Streamed intermediate results (`FunctionCallResultProperties(is_final=False)`) are still not supported on any of these realtime services. `GeminiLiveLLMService` and `InworldRealtimeLLMService` are excluded for now: Gemini Live's async-tool path needs deeper investigation, and Inworld appears to have a pre-existing problem with even simple tool calling on its Realtime API.

examples/realtime/realtime-azure-async-tool.py

@@ -0,0 +1,195 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example: async function call with the Azure Realtime LLM service.
+
+The ``get_current_weather`` tool is registered with
+``cancel_on_interruption=False`` and simulates a slow API call (10s sleep).
+While the call is in flight the conversation continues; the result arrives
+later via the async-tool mechanism and is forwarded to Azure Realtime as a
+``function_call_output`` so the model can integrate it naturally into its
+next turn.
+"""
+
+import asyncio
+import os
+import random
+from datetime import datetime
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.azure.realtime.llm import AzureRealtimeLLMService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.realtime.events import (
+    AudioConfiguration,
+    AudioInput,
+    InputAudioTranscription,
+    SessionProperties,
+)
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+async def fetch_weather_from_api(params: FunctionCallParams):
+    # Simulate a long-running API call so we can demonstrate that the
+    # conversation continues while the tool is in flight.
+    await asyncio.sleep(10)
+    temperature = (
+        random.randint(60, 85)
+        if params.arguments["format"] == "fahrenheit"
+        else random.randint(15, 30)
+    )
+    await params.result_callback(
+        {
+            "conditions": "nice",
+            "temperature": temperature,
+            "location": params.arguments["location"],
+            "format": params.arguments["format"],
+            "timestamp": datetime.now().strftime("%Y%m%d_%H%M%S"),
+        }
+    )
+
+
+weather_function = FunctionSchema(
+    name="get_current_weather",
+    description="Get the current weather",
+    properties={
+        "location": {
+            "type": "string",
+            "description": "The city and state, e.g. San Francisco, CA",
+        },
+        "format": {
+            "type": "string",
+            "enum": ["celsius", "fahrenheit"],
+            "description": "The temperature unit to use. Infer this from the users location.",
+        },
+    },
+    required=["location", "format"],
+)
+
+tools = ToolsSchema(standard_tools=[weather_function])
+
+
+system_instruction = (
+    "You are a friendly assistant. The user and you will engage in a spoken "
+    "dialog exchanging the transcripts of a natural real-time conversation. "
+    "Keep your responses short, generally two or three sentences for chatty "
+    "scenarios. When the user asks for the weather, call get_current_weather. "
+    "While you wait for the result, keep chatting with the user. When the "
+    "result arrives, share it with the user naturally."
+)
+
+
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    llm = AzureRealtimeLLMService(
+        api_key=os.environ["AZURE_REALTIME_API_KEY"],
+        base_url=os.environ["AZURE_REALTIME_BASE_URL"],
+        settings=AzureRealtimeLLMService.Settings(
+            system_instruction=system_instruction,
+            session_properties=SessionProperties(
+                audio=AudioConfiguration(
+                    input=AudioInput(
+                        transcription=InputAudioTranscription(model="whisper-1"),
+                    )
+                ),
+            ),
+        ),
+    )
+
+    llm.register_function(
+        "get_current_weather",
+        fetch_weather_from_api,
+        cancel_on_interruption=False,
+    )
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            user_aggregator,
+            llm,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/realtime/realtime-grok-async-tool.py

@@ -0,0 +1,179 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example: async function call with the Grok Realtime LLM service.
+
+The ``get_current_weather`` tool is registered with
+``cancel_on_interruption=False`` and simulates a slow API call (10s sleep).
+While the call is in flight the conversation continues; the result arrives
+later via the async-tool mechanism and is forwarded to Grok Realtime as a
+``function_call_output`` so the model can integrate it naturally into its
+next turn.
+"""
+
+import asyncio
+import os
+import random
+from datetime import datetime
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import LLMContextAggregatorPair
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.xai.realtime.events import SessionProperties
+from pipecat.services.xai.realtime.llm import GrokRealtimeLLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+async def fetch_weather_from_api(params: FunctionCallParams):
+    # Simulate a long-running API call so we can demonstrate that the
+    # conversation continues while the tool is in flight.
+    await asyncio.sleep(10)
+    temperature = (
+        random.randint(60, 85)
+        if params.arguments["format"] == "fahrenheit"
+        else random.randint(15, 30)
+    )
+    await params.result_callback(
+        {
+            "conditions": "nice",
+            "temperature": temperature,
+            "location": params.arguments["location"],
+            "format": params.arguments["format"],
+            "timestamp": datetime.now().strftime("%Y%m%d_%H%M%S"),
+        }
+    )
+
+
+weather_function = FunctionSchema(
+    name="get_current_weather",
+    description="Get the current weather",
+    properties={
+        "location": {
+            "type": "string",
+            "description": "The city and state, e.g. San Francisco, CA",
+        },
+        "format": {
+            "type": "string",
+            "enum": ["celsius", "fahrenheit"],
+            "description": "The temperature unit to use. Infer this from the users location.",
+        },
+    },
+    required=["location", "format"],
+)
+
+tools = ToolsSchema(standard_tools=[weather_function])
+
+
+system_instruction = (
+    "You are a friendly assistant. The user and you will engage in a spoken "
+    "dialog exchanging the transcripts of a natural real-time conversation. "
+    "Keep your responses short, generally two or three sentences for chatty "
+    "scenarios. When the user asks for the weather, call get_current_weather. "
+    "While you wait for the result, keep chatting with the user. When the "
+    "result arrives, share it with the user naturally."
+)
+
+
+# Note: Grok has built-in server-side VAD, so we don't need local VAD.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    llm = GrokRealtimeLLMService(
+        api_key=os.environ["XAI_API_KEY"],
+        settings=GrokRealtimeLLMService.Settings(
+            system_instruction=system_instruction,
+            session_properties=SessionProperties(
+                voice="Ara",
+            ),
+        ),
+    )
+
+    llm.register_function(
+        "get_current_weather",
+        fetch_weather_from_api,
+        cancel_on_interruption=False,
+    )
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(context)
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            user_aggregator,
+            llm,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/realtime/realtime-ultravox-async-tool.py

@@ -0,0 +1,186 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Example: async function call with the Ultravox Realtime LLM service.
+
+The ``get_current_weather`` tool is registered with
+``cancel_on_interruption=False`` and simulates a slow API call (10s sleep).
+
+Ultravox's API freezes the conversation between ``client_tool_invocation``
+and the matching ``client_tool_result``, so the service ships a placeholder
+``client_tool_result`` immediately when an async-registered function is
+invoked (to unfreeze the conversation). When the real tool finishes, the
+actual result is injected as user-side text so the model picks it up.
+"""
+
+import asyncio
+import datetime
+import os
+import random
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.ultravox.llm import OneShotInputParams, UltravoxRealtimeLLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_stop import SpeechTimeoutUserTurnStopStrategy
+from pipecat.turns.user_turn_strategies import UserTurnStrategies
+
+load_dotenv(override=True)
+
+
+async def fetch_weather_from_api(params: FunctionCallParams):
+    # Simulate a long-running API call so we can demonstrate that the
+    # conversation continues while the tool is in flight.
+    await asyncio.sleep(10)
+    temperature = (
+        random.randint(60, 85)
+        if params.arguments["format"] == "fahrenheit"
+        else random.randint(15, 30)
+    )
+    await params.result_callback(
+        {
+            "conditions": "nice",
+            "temperature": temperature,
+            "location": params.arguments["location"],
+            "format": params.arguments["format"],
+            "timestamp": datetime.datetime.now().strftime("%Y%m%d_%H%M%S"),
+        }
+    )
+
+
+weather_function = FunctionSchema(
+    name="get_current_weather",
+    description="Get the current weather",
+    properties={
+        "location": {
+            "type": "string",
+            "description": "The city and state, e.g. San Francisco, CA",
+        },
+        "format": {
+            "type": "string",
+            "enum": ["celsius", "fahrenheit"],
+            "description": "The temperature unit to use. Infer this from the users location.",
+        },
+    },
+    required=["location", "format"],
+)
+
+
+system_prompt = (
+    "You are a friendly assistant. The user and you will engage in a spoken "
+    "dialog exchanging the transcripts of a natural real-time conversation. "
+    "Keep your responses short, generally two or three sentences for chatty "
+    "scenarios. When the user asks for the weather, call get_current_weather. "
+    "While you wait for the result, keep chatting with the user. When the "
+    "result arrives, share it with the user naturally."
+)
+
+
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    llm = UltravoxRealtimeLLMService(
+        params=OneShotInputParams(
+            api_key=os.environ["ULTRAVOX_API_KEY"],
+            system_prompt=system_prompt,
+            temperature=0.3,
+            max_duration=datetime.timedelta(minutes=3),
+        ),
+        one_shot_selected_tools=ToolsSchema(standard_tools=[weather_function]),
+    )
+
+    llm.register_function(
+        "get_current_weather",
+        fetch_weather_from_api,
+        cancel_on_interruption=False,
+    )
+
+    context = LLMContext([])
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            user_turn_strategies=UserTurnStrategies(
+                stop=[SpeechTimeoutUserTurnStopStrategy()],
+            ),
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            user_aggregator,
+            llm,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

src/pipecat/services/llm_service.py

@@ -751,6 +751,19 @@ class LLMService(UserTurnCompletionLLMServiceMixin, AIService, Generic[TAdapter]
             return True
         return function_name in self._functions.keys()
 
+    def _function_is_async(self, function_name: str) -> bool:
+        """Whether the named function was registered with cancel_on_interruption=False.
+
+        Mirrors the registry-lookup pattern in :meth:`run_function_calls`:
+        a name-specific entry takes precedence; if there isn't one, fall
+        back to the ``None``-keyed catch-all entry. Returns ``False`` if
+        no entry matches.
+        """
+        item = self._functions.get(function_name)
+        if item is None:
+            item = self._functions.get(None)
+        return item is not None and not item.cancel_on_interruption
+
     async def run_function_calls(self, function_calls: Sequence[FunctionCallFromLLM]):
         """Execute a sequence of function calls from the LLM.
 

src/pipecat/services/ultravox/llm.py

@@ -45,7 +45,8 @@ from pipecat.frames.frames import (
     UserAudioRawFrame,
     VADUserStoppedSpeakingFrame,
 )
-from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators import async_tool_messages
+from pipecat.processors.aggregators.llm_context import LLMContext, LLMSpecificMessage
 from pipecat.processors.frame_processor import FrameDirection
 from pipecat.services.llm_service import FunctionCallFromLLM, LLMService
 from pipecat.services.settings import NOT_GIVEN, LLMSettings, _NotGiven, assert_given
@@ -59,6 +60,24 @@ except ModuleNotFoundError as e:
     raise Exception(f"Missing module: {e}")
 
 
+# Result shipped as the client_tool_result when we see an async-tool
+# "started" message — i.e. when an async-registered function call
+# (cancel_on_interruption=False) is invoked. Sending it immediately
+# unfreezes the conversation so the model can keep talking while the
+# real tool runs; the actual result is injected later as user-side text
+# once the tool finishes.
+_ASYNC_TOOL_STARTED_RESULT = (
+    "The actual result for this tool call is not yet ready. A follow-up "
+    "message will arrive shortly with the actual result. In the meantime, "
+    "keep the conversation going naturally."
+)
+
+# Template for the user-side text we inject when the async-tool "final"
+# message arrives. Bracketed framing helps the model treat this as a
+# tool-result update rather than fresh user input.
+_ASYNC_TOOL_FINAL_RESULT_TEMPLATE = "[Async tool result for tool_call_id={tool_call_id}] {result}"
+
+
 @dataclass
 class UltravoxRealtimeLLMSettings(LLMSettings):
     """Settings for UltravoxRealtimeLLMService.
@@ -218,6 +237,11 @@ class UltravoxRealtimeLLMService(LLMService):
         self._disconnecting = False
         self._bot_responding: Literal[None, "text", "voice"] = None
         self._last_user_id: str | None = None
+        self._completed_tool_calls: set[str] = set()
+        # Tracks tool_call_ids for which we've already shipped the
+        # async-tool placeholder client_tool_result that unfreezes the
+        # conversation while the real tool runs. See _handle_tool_invocation.
+        self._started_placeholder_sent: set[str] = set()
 
         self._sample_rate = 48000
         self._resampler = create_stream_resampler()
@@ -373,6 +397,8 @@ class UltravoxRealtimeLLMService(LLMService):
         if self._receive_task:
             await self.cancel_task(self._receive_task, timeout=1.0)
             self._receive_task = None
+        self._completed_tool_calls = set()
+        self._started_placeholder_sent = set()
 
     async def _update_settings(self, delta: Settings):
         changed = await super()._update_settings(delta)
@@ -413,20 +439,80 @@ class UltravoxRealtimeLLMService(LLMService):
             await self.push_frame(frame, direction)
 
     async def _handle_context(self, context: LLMContext):
-        # Ultravox handles all context server-side, so the only context we may
+        # Ultravox handles all context server-side, so the only context we
-        # need to handle here is new function call results.
+        # need to handle here is function-call results.
-        for message in reversed(context.messages):
+        for message in context.get_messages():
-            if message.get("role") != "tool":
+            # LLMSpecificMessages are opaque provider-specific payloads, not
-                break
+            # standard tool-result messages — skip them.
-            content = message.get("content")
+            if isinstance(message, LLMSpecificMessage):
-            socket_message = {
+                continue
+
+            # Async-tool messages live alongside regular tool messages in the
+            # context; detect and route them before the regular logic so we
+            # don't try to send the async-tool envelope JSON as a tool result.
+            async_payload = async_tool_messages.parse_message(message)
+            if async_payload is not None:
+                if async_payload.kind == "started":
+                    # The placeholder client_tool_result that unfreezes the
+                    # conversation was already shipped from
+                    # _handle_tool_invocation when the model issued the
+                    # call. Nothing more to do here.
+                    continue
+                if async_payload.kind == "intermediate":
+                    logger.error(
+                        f"{self}: Ultravox does not support streamed async "
+                        f"tool results; dropping intermediate result for "
+                        f"tool_call_id={async_payload.tool_call_id}. Use a "
+                        f"non-realtime LLM service if your tool needs to "
+                        f"stream intermediate results."
+                    )
+                    await self.push_error(
+                        error_msg="Ultravox does not support streamed async tool results.",
+                    )
+                    continue
+                if async_payload.kind == "final":
+                    if async_payload.tool_call_id in self._completed_tool_calls:
+                        continue
+                    # The placeholder client_tool_result has already
+                    # "completed" the tool call from Ultravox's perspective,
+                    # so the actual result is delivered as user-side text
+                    # (see _ASYNC_TOOL_FINAL_RESULT_TEMPLATE).
+                    await self._send_user_text(
+                        _ASYNC_TOOL_FINAL_RESULT_TEMPLATE.format(
+                            tool_call_id=async_payload.tool_call_id,
+                            result=async_payload.result,
+                        )
+                    )
+                    self._completed_tool_calls.add(async_payload.tool_call_id)
+                    continue
+                # Defensive: any async-tool message must not fall through
+                # to the regular tool-result block below, even if it
+                # carries a kind we don't recognize.
+                continue
+
+            # Look for newly-completed "regular" (as opposed to async-tool) results
+            if message.get("role") == "tool" and message.get("content") != "IN_PROGRESS":
+                tool_call_id = message.get("tool_call_id")
+                if tool_call_id and tool_call_id not in self._completed_tool_calls:
+                    content = message.get("content")
+                    result = (
+                        content
+                        if isinstance(content, str)
+                        else "".join(t.get("text") for t in content)
+                    )
+                    await self._send_tool_result(tool_call_id, result)
+                    self._completed_tool_calls.add(tool_call_id)
+
+    async def _send_tool_result(self, tool_call_id: str, result: str):
+        """Send a tool call result to Ultravox."""
+        logger.debug(f"Sending tool result to Ultravox for tool_call_id={tool_call_id}")
+        await self._send(
+            {
                 "type": "client_tool_result",
-                "invocationId": message.get("tool_call_id"),
+                "invocationId": tool_call_id,
-                "result": content
+                "result": result,
-                if isinstance(content, str)
-                else "".join(t.get("text") for t in content),
             }
-            await self._send(socket_message)
+        )
 
     async def _handle_vad_user_stopped_speaking(self, frame: VADUserStoppedSpeakingFrame):
         """Handle VAD user stopped speaking frame.
@@ -567,6 +653,19 @@ class UltravoxRealtimeLLMService(LLMService):
     async def _handle_tool_invocation(
         self, tool_name: str, invocation_id: str, parameters: dict[str, Any]
     ):
+        # Ultravox freezes the conversation between client_tool_invocation
+        # and the matching client_tool_result. For functions registered
+        # with cancel_on_interruption=False the actual result won't be
+        # available for some time, so ship a placeholder result now to
+        # unfreeze the conversation. The real result will be injected
+        # later as user-side text from _handle_context.
+        if (
+            self._function_is_async(tool_name)
+            and invocation_id not in self._started_placeholder_sent
+        ):
+            await self._send_tool_result(invocation_id, _ASYNC_TOOL_STARTED_RESULT)
+            self._started_placeholder_sent.add(invocation_id)
+
         await self.run_function_calls(
             [
                 FunctionCallFromLLM(

src/pipecat/services/xai/realtime/llm.py

@@ -47,7 +47,8 @@ from pipecat.frames.frames import (
     UserStoppedSpeakingFrame,
 )
 from pipecat.metrics.metrics import LLMTokenUsage
-from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators import async_tool_messages
+from pipecat.processors.aggregators.llm_context import LLMContext, LLMSpecificMessage
 from pipecat.processors.frame_processor import FrameDirection
 from pipecat.services.llm_service import FunctionCallFromLLM, LLMService
 from pipecat.services.settings import (
@@ -913,6 +914,50 @@ class GrokRealtimeLLMService(LLMService[GrokRealtimeLLMAdapter]):
         sent_new_result = False
 
         for message in self._context.get_messages():
+            # LLMSpecificMessages are opaque provider-specific payloads, not
+            # standard tool-result messages — skip them.
+            if isinstance(message, LLMSpecificMessage):
+                continue
+
+            # Async-tool messages live alongside regular tool messages in the
+            # context; detect and route them before the regular logic so we
+            # don't try to send the async-tool envelope JSON as a tool result.
+            async_payload = async_tool_messages.parse_message(message)
+            if async_payload is not None:
+                if async_payload.tool_call_id in self._completed_tool_calls:
+                    continue
+                if async_payload.kind == "started":
+                    # The provider already issued the tool call and natively
+                    # awaits a result; nothing to send for the started marker.
+                    continue
+                if async_payload.kind == "intermediate":
+                    logger.error(
+                        f"{self}: Grok Realtime does not support streamed async "
+                        f"tool results; dropping intermediate result for "
+                        f"tool_call_id={async_payload.tool_call_id}. Use a "
+                        f"non-realtime LLM service if your tool needs to "
+                        f"stream intermediate results."
+                    )
+                    await self.push_error(
+                        error_msg="Grok Realtime does not support streamed async tool results.",
+                    )
+                    continue
+                if async_payload.kind == "final":
+                    # Deliver via the formal tool-result channel — same path
+                    # as a synchronous tool result, just delayed.
+                    if send_new_results:
+                        sent_new_result = True
+                        await self._send_tool_result(
+                            async_payload.tool_call_id, async_payload.result
+                        )
+                    self._completed_tool_calls.add(async_payload.tool_call_id)
+                    continue
+                # Defensive: any async-tool message must not fall through
+                # to the regular tool-result block below, even if it
+                # carries a kind we don't recognize.
+                continue
+
+            # Look for newly-completed "regular" (as opposed to async-tool) results
             if message.get("role") and message.get("content") != "IN_PROGRESS":
                 tool_call_id = message.get("tool_call_id")
                 if tool_call_id and tool_call_id not in self._completed_tool_calls:
@@ -939,6 +984,7 @@ class GrokRealtimeLLMService(LLMService[GrokRealtimeLLMAdapter]):
 
     async def _send_tool_result(self, tool_call_id: str, result: str):
         """Send a tool call result to Grok."""
+        logger.debug(f"Sending tool result to Grok Realtime for tool_call_id={tool_call_id}")
         item = events.ConversationItem(
             type="function_call_output",
             call_id=tool_call_id,