Remove error-wrapping overrides and align class docstring

The OpenAI SDK already preserves server error details in typed exceptions
(APIStatusError subclasses), and BaseOpenAILLMService propagates errors
via push_error(). The RuntimeError wrapper was erasing exception types
that callers may want to handle explicitly.

examples/foundational/14y-function-calling-sarvam.py

@@ -0,0 +1,178 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+
+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, TTSSpeakFrame
+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.sarvam.llm import SarvamLLMService
+from pipecat.services.sarvam.stt import SarvamSTTService
+from pipecat.services.sarvam.tts import SarvamTTSService
+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):
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+async def fetch_restaurant_recommendation(params: FunctionCallParams):
+    await params.result_callback({"name": "The Golden Dragon"})
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+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")
+
+    stt = SarvamSTTService(
+        api_key=os.getenv("SARVAM_API_KEY"),
+        settings=SarvamSTTService.Settings(
+            model="saarika:v2.5",
+        ),
+    )
+
+    tts = SarvamTTSService(
+        api_key=os.getenv("SARVAM_API_KEY"),
+        settings=SarvamTTSService.Settings(
+            model="bulbul:v2",
+            voice="manisha",
+        ),
+    )
+    llm = SarvamLLMService(
+        api_key=os.getenv("SARVAM_API_KEY"),
+        settings=SarvamLLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    # You can also register a function_name of None to get all functions
+    # sent to the same callback with an additional function_name parameter.
+    llm.register_function("get_current_weather", fetch_weather_from_api)
+    llm.register_function("get_restaurant_recommendation", fetch_restaurant_recommendation)
+
+    @llm.event_handler("on_function_calls_started")
+    async def on_function_calls_started(service, function_calls):
+        await tts.queue_frame(TTSSpeakFrame("Let me check on that."))
+
+    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 user's location.",
+            },
+        },
+        required=["location", "format"],
+    )
+    restaurant_function = FunctionSchema(
+        name="get_restaurant_recommendation",
+        description="Get a restaurant recommendation",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+        },
+        required=["location"],
+    )
+    tools = ToolsSchema(standard_tools=[weather_function, restaurant_function])
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            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")
+        # Kick off the conversation.
+        context.add_message({"role": "user", "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()

scripts/evals/run-release-evals.py

@@ -186,6 +186,7 @@ TESTS_14 = [
     ("14v-function-calling-openai.py", EVAL_WEATHER),
     ("14w-function-calling-mistral.py", EVAL_WEATHER),
     ("14x-function-calling-openpipe.py", EVAL_WEATHER),
+    ("14y-function-calling-sarvam.py", EVAL_WEATHER),
     ("14-function-calling-openai-responses.py", EVAL_WEATHER),
     ("14-function-calling-openai-responses.py", EVAL_WEATHER_AND_RESTAURANT),
     # Video

src/pipecat/services/sarvam/llm.py

@@ -6,27 +6,19 @@
 
 """Sarvam LLM service implementation using OpenAI-compatible interface."""
 
-import asyncio
-import json
 from dataclasses import dataclass, field
-from typing import Any, Awaitable, Literal, Mapping, Optional, TypeVar
+from typing import Literal, Mapping, Optional
 
-import httpx
 from loguru import logger
-from openai import NOT_GIVEN, APITimeoutError, AsyncStream
+from openai import NOT_GIVEN
-from openai.types.chat import ChatCompletionChunk
 
 from pipecat.adapters.services.open_ai_adapter import OpenAILLMInvocationParams
-from pipecat.processors.aggregators.llm_context import LLMContext
-from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
 from pipecat.services.openai.base_llm import OpenAILLMSettings
 from pipecat.services.openai.llm import OpenAILLMService
 from pipecat.services.sarvam._sdk import sdk_headers
 from pipecat.services.settings import NOT_GIVEN as _NOT_GIVEN
 from pipecat.services.settings import _NotGiven, is_given
 
-_T = TypeVar("_T")
-
 
 @dataclass
 class SarvamLLMSettings(OpenAILLMSettings):
@@ -44,15 +36,10 @@ class SarvamLLMSettings(OpenAILLMSettings):
 
 
 class SarvamLLMService(OpenAILLMService):
-    """Sarvam LLM service using Sarvam's OpenAI-compatible chat completions API.
+    """A service for interacting with Sarvam's API using the OpenAI-compatible interface.
 
-    This service extends ``OpenAILLMService`` while adding Sarvam-specific behavior:
+    This service extends OpenAILLMService to connect to Sarvam's API endpoint while
-
+    maintaining full compatibility with OpenAI's interface and functionality.
-    - model allow-list validation
-    - request shaping for Sarvam-compatible parameters
-    - Sarvam auth header wiring (``api-subscription-key``)
-    - SDK User-Agent propagation on every API call
-    - raw Sarvam server error passthrough
     """
 
     _SUPPORTED_MODELS = frozenset(
@@ -153,44 +140,6 @@ class SarvamLLMService(OpenAILLMService):
 
         return params
 
-    async def _call_with_raw_sarvam_errors(self, awaitable: Awaitable[_T]) -> _T:
-        """Await an OpenAI call while preserving Sarvam raw error payloads.
-
-        BaseOpenAILLMService handles pipeline-frame exceptions via push_error(),
-        but direct helper methods like ``get_chat_completions`` and
-        ``run_inference`` are often consumed directly. We normalize those errors
-        here so applications consistently receive server-provided messages.
-        """
-        try:
-            return await awaitable
-        except (APITimeoutError, asyncio.TimeoutError, httpx.TimeoutException):
-            raise
-        except Exception as e:
-            raise RuntimeError(self._format_raw_server_error(e)) from e
-
-    async def get_chat_completions(
-        self, params_from_context: OpenAILLMInvocationParams
-    ) -> AsyncStream[ChatCompletionChunk]:
-        """Get streaming chat completions with Sarvam raw error passthrough."""
-        return await self._call_with_raw_sarvam_errors(
-            super().get_chat_completions(params_from_context)
-        )
-
-    async def run_inference(
-        self,
-        context: LLMContext | OpenAILLMContext,
-        max_tokens: Optional[int] = None,
-        system_instruction: Optional[str] = None,
-    ) -> Optional[str]:
-        """Run one-shot inference and preserve Sarvam raw server errors."""
-        return await self._call_with_raw_sarvam_errors(
-            super().run_inference(
-                context,
-                max_tokens=max_tokens,
-                system_instruction=system_instruction,
-            )
-        )
-
     def _validate_model(self, model: str):
         if model not in self._SUPPORTED_MODELS:
             allowed = ", ".join(sorted(self._SUPPORTED_MODELS))
@@ -209,35 +158,3 @@ class SarvamLLMService(OpenAILLMService):
 
         if has_tool_choice and not has_tools:
             raise ValueError("Sarvam requires non-empty `tools` when `tool_choice` is provided.")
-
-    def _format_raw_server_error(self, error: Exception) -> str:
-        raw_message = self._extract_raw_server_message(error)
-        return f"Sarvam server error: {raw_message}"
-
-    def _extract_raw_server_message(self, error: Exception) -> str:
-        body = getattr(error, "body", None)
-        if body is not None:
-            return self._payload_to_message(body)
-
-        response = getattr(error, "response", None)
-        if response is not None:
-            try:
-                return self._payload_to_message(response.json())
-            except Exception:
-                text = getattr(response, "text", None)
-                if text:
-                    return str(text)
-
-        return str(error)
-
-    def _payload_to_message(self, payload: Any) -> str:
-        if isinstance(payload, dict):
-            error_obj = payload.get("error")
-            if isinstance(error_obj, dict) and isinstance(error_obj.get("message"), str):
-                return error_obj["message"]
-            if isinstance(payload.get("message"), str):
-                return payload["message"]
-            return json.dumps(payload, ensure_ascii=False)
-        if isinstance(payload, list):
-            return json.dumps(payload, ensure_ascii=False)
-        return str(payload)