From 2b286365e0f2a3f9f764f0ebdbbd586eaa782abf Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 10:44:51 -0400
Subject: [PATCH 01/23] Add MISTRAL_API_KEY to env.example

---
 env.example | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/env.example b/env.example
index 2962acc11..a6ede22d7 100644
--- a/env.example
+++ b/env.example
@@ -146,3 +146,6 @@ SENTRY_DSN=...
 
 # Heygen
 HEYGEN_API_KEY=...
+
+# Mistral
+MISTRAL_API_KEY=...

From b22bdff6d078cdf6355ad3acad94e63d022be854 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 10:57:46 -0400
Subject: [PATCH 02/23] Add support for universal `LLMContext` to Azure LLM
 service

---
 src/pipecat/services/azure/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/azure/llm.py b/src/pipecat/services/azure/llm.py
index 47a6ef280..a4b93f2a4 100644
--- a/src/pipecat/services/azure/llm.py
+++ b/src/pipecat/services/azure/llm.py
@@ -60,12 +60,3 @@ class AzureLLMService(OpenAILLMService):
             azure_endpoint=self._endpoint,
             api_version=self._api_version,
         )
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as Azure service does yet not support universal LLMContext.
-        """
-        return False

From d45fbd5455288cc44bb2469a2dfed61d3f53d09d Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 11:17:01 -0400
Subject: [PATCH 03/23] Add support for universal `LLMContext` to Cerebras LLM
 service

---
 src/pipecat/services/cerebras/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/cerebras/llm.py b/src/pipecat/services/cerebras/llm.py
index 9bdc5b963..5d6690d73 100644
--- a/src/pipecat/services/cerebras/llm.py
+++ b/src/pipecat/services/cerebras/llm.py
@@ -81,12 +81,3 @@ class CerebrasLLMService(OpenAILLMService):
 
         params.update(self._settings["extra"])
         return params
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as Cerebras service does not yet support universal LLMContext.
-        """
-        return False

From 2dbd17da4df3b3ef794f4830b4b444c61e9ca767 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 12:20:40 -0400
Subject: [PATCH 04/23] Fix Cerebras function calling example

---
 examples/foundational/14k-function-calling-cerebras.py | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/examples/foundational/14k-function-calling-cerebras.py b/examples/foundational/14k-function-calling-cerebras.py
index b96a979c5..bb89893b9 100644
--- a/examples/foundational/14k-function-calling-cerebras.py
+++ b/examples/foundational/14k-function-calling-cerebras.py
@@ -72,9 +72,13 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     # sent to the same callback with an additional function_name parameter.
     llm.register_function("get_current_weather", fetch_weather_from_api)
 
-    @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."))
+    # Disabling for now, as we end up in an infinite inference loop with the
+    # model in this example ("let me check on that" ends up at the end of the
+    # context, which the model erroneously treats as a nudge to call the tool
+    # again).
+    # @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",

From 2b00d37e949033079f0e6e1d2a56ab7850968cac Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 13:59:49 -0400
Subject: [PATCH 05/23] Add support for universal `LLMContext` to DeepSeek LLM
 service

---
 src/pipecat/services/deepseek/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/deepseek/llm.py b/src/pipecat/services/deepseek/llm.py
index 7b616a293..7414a1f33 100644
--- a/src/pipecat/services/deepseek/llm.py
+++ b/src/pipecat/services/deepseek/llm.py
@@ -82,12 +82,3 @@ class DeepSeekLLMService(OpenAILLMService):
 
         params.update(self._settings["extra"])
         return params
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as DeepSeekLLMService does not yet support universal LLMContext.
-        """
-        return False

From f8e13a82cfbb888b5f7bba5a1e442ae0f5e8c6c0 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 14:22:33 -0400
Subject: [PATCH 06/23] Fix Fireworks AI function calling example

---
 .../foundational/14i-function-calling-fireworks.py   | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/examples/foundational/14i-function-calling-fireworks.py b/examples/foundational/14i-function-calling-fireworks.py
index 4adc65525..1f1f5c692 100644
--- a/examples/foundational/14i-function-calling-fireworks.py
+++ b/examples/foundational/14i-function-calling-fireworks.py
@@ -75,9 +75,13 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     # sent to the same callback with an additional function_name parameter.
     llm.register_function("get_current_weather", fetch_weather_from_api)
 
-    @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."))
+    # Disabling for now, as it ends up tripping up the model in this example
+    # ("let me check on that" ends up at the end of the context, which the
+    # model erroneously treats as a nudge to call the tool again; the
+    # ensuing inference then yields wonky results).
+    # @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",
@@ -99,7 +103,7 @@ async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
     messages = [
         {
             "role": "system",
-            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio so don't include special characters in your answers. Respond to what the user said in a creative and helpful way.",
+            "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio so don't include special characters in your answers. Respond to what the user said in a creative and helpful way. Start by saying hello.",
         },
     ]
 

From da18d0a5625528059f319c55c19d3de9028154fb Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 14:24:22 -0400
Subject: [PATCH 07/23] Add support for universal `LLMContext` to Fireworks AI
 LLM service

---
 src/pipecat/services/fireworks/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/fireworks/llm.py b/src/pipecat/services/fireworks/llm.py
index 194adfc51..29c3f0284 100644
--- a/src/pipecat/services/fireworks/llm.py
+++ b/src/pipecat/services/fireworks/llm.py
@@ -82,12 +82,3 @@ class FireworksLLMService(OpenAILLMService):
 
         params.update(self._settings["extra"])
         return params
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as FireworksLLMService does not yet support universal LLMContext.
-        """
-        return False

From d8882bc381399ade2b755f3d627205e304f41c1b Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 14:38:44 -0400
Subject: [PATCH 08/23] Add support for universal `LLMContext` to Google Vertex
 AI LLM service

---
 src/pipecat/services/google/llm_vertex.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/google/llm_vertex.py b/src/pipecat/services/google/llm_vertex.py
index bdbf2dda1..22b6258a5 100644
--- a/src/pipecat/services/google/llm_vertex.py
+++ b/src/pipecat/services/google/llm_vertex.py
@@ -139,12 +139,3 @@ class GoogleVertexLLMService(OpenAILLMService):
         creds.refresh(Request())  # Ensure token is up-to-date, lifetime is 1 hour.
 
         return creds.token
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as GoogleVertexLLMService does not yet support universal LLMContext.
-        """
-        return False

From b4e105106610ff3493eb6584e1ec7a556a65dd64 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 14:42:46 -0400
Subject: [PATCH 09/23] Add support for universal `LLMContext` to Grok LLM
 service

---
 src/pipecat/services/grok/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/grok/llm.py b/src/pipecat/services/grok/llm.py
index 49fe2e802..2a9704008 100644
--- a/src/pipecat/services/grok/llm.py
+++ b/src/pipecat/services/grok/llm.py
@@ -190,12 +190,3 @@ class GrokLLMService(OpenAILLMService):
         user = OpenAIUserContextAggregator(context, params=user_params)
         assistant = OpenAIAssistantContextAggregator(context, params=assistant_params)
         return GrokContextAggregatorPair(_user=user, _assistant=assistant)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as GrokLLMService does not yet support universal LLMContext.
-        """
-        return False

From 34f823bcacc70547a502b33a2e8594ca18a4ebab Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 14:49:14 -0400
Subject: [PATCH 10/23] Add support for universal `LLMContext` to Groq LLM
 service

---
 src/pipecat/services/groq/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/groq/llm.py b/src/pipecat/services/groq/llm.py
index d3166ff8b..57f2a533d 100644
--- a/src/pipecat/services/groq/llm.py
+++ b/src/pipecat/services/groq/llm.py
@@ -49,12 +49,3 @@ class GroqLLMService(OpenAILLMService):
         """
         logger.debug(f"Creating Groq client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as GroqLLMService does not yet support universal LLMContext.
-        """
-        return False

From ac62183eb61fa72be6b69d655e7b5da12f2b0224 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 14:52:09 -0400
Subject: [PATCH 11/23] Add NVIDIA_API_KEY to env.example

---
 env.example | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/env.example b/env.example
index a6ede22d7..bb8114bae 100644
--- a/env.example
+++ b/env.example
@@ -149,3 +149,6 @@ HEYGEN_API_KEY=...
 
 # Mistral
 MISTRAL_API_KEY=...
+
+# NVIDIA
+NVIDIA_API_KEY=

From d740bab99e8dcb4fc16f5bdf9595408be8062d60 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:00:01 -0400
Subject: [PATCH 12/23] Add support for universal `LLMContext` to NVIDIA NIM
 LLM service

---
 src/pipecat/services/nim/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/nim/llm.py b/src/pipecat/services/nim/llm.py
index fdfb8bf6b..052b94274 100644
--- a/src/pipecat/services/nim/llm.py
+++ b/src/pipecat/services/nim/llm.py
@@ -47,15 +47,6 @@ class NimLLMService(OpenAILLMService):
         self._has_reported_prompt_tokens = False
         self._is_processing = False
 
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as NimLLMService does not yet support universal LLMContext.
-        """
-        return False
-
     async def _process_context(self, context: OpenAILLMContext):
         """Process a context through the LLM and accumulate token usage metrics.
 

From 0c32cc29a719a1341f981e5cff27b579146f7712 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:13:14 -0400
Subject: [PATCH 13/23] Add support for universal `LLMContext` to OpenPipe LLM
 service

---
 src/pipecat/services/openpipe/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/openpipe/llm.py b/src/pipecat/services/openpipe/llm.py
index 2e491ea26..ad85e8568 100644
--- a/src/pipecat/services/openpipe/llm.py
+++ b/src/pipecat/services/openpipe/llm.py
@@ -108,12 +108,3 @@ class OpenPipeLLMService(OpenAILLMService):
         }
 
         return params
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as OpenPipeLLMService does not yet support universal LLMContext.
-        """
-        return False

From ec138b97d9bc1748ae6b256364041db62d17e05b Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:19:32 -0400
Subject: [PATCH 14/23] Add support for universal `LLMContext` to OpenRouter
 LLM service

---
 src/pipecat/services/openrouter/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/openrouter/llm.py b/src/pipecat/services/openrouter/llm.py
index 3ba1ae6f6..97a9d336a 100644
--- a/src/pipecat/services/openrouter/llm.py
+++ b/src/pipecat/services/openrouter/llm.py
@@ -61,12 +61,3 @@ class OpenRouterLLMService(OpenAILLMService):
         """
         logger.debug(f"Creating OpenRouter client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as OpenRouterLLMService does not yet support universal LLMContext.
-        """
-        return False

From 5bc0e49baa4a784ad1cada4600ecfe22e73b0d2c Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:23:35 -0400
Subject: [PATCH 15/23] Add support for universal `LLMContext` to Perplexity
 LLM service

---
 src/pipecat/services/perplexity/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/perplexity/llm.py b/src/pipecat/services/perplexity/llm.py
index 3e39206c6..bc4b220ac 100644
--- a/src/pipecat/services/perplexity/llm.py
+++ b/src/pipecat/services/perplexity/llm.py
@@ -84,15 +84,6 @@ class PerplexityLLMService(OpenAILLMService):
 
         return params
 
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as PerplexityLLMService does not yet support universal LLMContext.
-        """
-        return False
-
     async def _process_context(self, context: OpenAILLMContext):
         """Process a context through the LLM and accumulate token usage metrics.
 

From d9ea02595b16af6060f8f31f8a44fedc5f81c1bc Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:28:02 -0400
Subject: [PATCH 16/23] Add support for universal `LLMContext` to Ollama LLM
 service

---
 src/pipecat/services/ollama/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/ollama/llm.py b/src/pipecat/services/ollama/llm.py
index aa6f58b59..2284a5070 100644
--- a/src/pipecat/services/ollama/llm.py
+++ b/src/pipecat/services/ollama/llm.py
@@ -43,12 +43,3 @@ class OLLamaLLMService(OpenAILLMService):
         """
         logger.debug(f"Creating Ollama client with api {base_url}")
         return super().create_client(base_url=base_url, **kwargs)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as OLLamaLLMService does not yet support universal LLMContext.
-        """
-        return False

From b24eb76559a5b75ef595d3e376da343657b18a28 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:29:22 -0400
Subject: [PATCH 17/23] Add QWEN_API_KEY to env.example

---
 env.example | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/env.example b/env.example
index bb8114bae..3690b5523 100644
--- a/env.example
+++ b/env.example
@@ -151,4 +151,7 @@ HEYGEN_API_KEY=...
 MISTRAL_API_KEY=...
 
 # NVIDIA
-NVIDIA_API_KEY=
+NVIDIA_API_KEY=...
+
+# Qwen
+QWEN_API_KEY=...

From 91a3f63e28aeb3fd01a3dac26bbd3d78d4ecf185 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:32:24 -0400
Subject: [PATCH 18/23] Add support for universal `LLMContext` to Qwen LLM
 service

---
 src/pipecat/services/qwen/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/qwen/llm.py b/src/pipecat/services/qwen/llm.py
index 1c842ded6..648cbd9e8 100644
--- a/src/pipecat/services/qwen/llm.py
+++ b/src/pipecat/services/qwen/llm.py
@@ -50,12 +50,3 @@ class QwenLLMService(OpenAILLMService):
         """
         logger.debug(f"Creating Qwen client with base URL: {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as QwenLLMService does not yet support universal LLMContext.
-        """
-        return False

From e04f42167e2e06e40de648422318223412590ee7 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:40:15 -0400
Subject: [PATCH 19/23] Add support for universal `LLMContext` to SambaNova LLM
 service

---
 src/pipecat/services/sambanova/llm.py | 22 +++++++++-------------
 1 file changed, 9 insertions(+), 13 deletions(-)

diff --git a/src/pipecat/services/sambanova/llm.py b/src/pipecat/services/sambanova/llm.py
index d39eb51a2..5ed600457 100644
--- a/src/pipecat/services/sambanova/llm.py
+++ b/src/pipecat/services/sambanova/llm.py
@@ -18,6 +18,7 @@ from pipecat.frames.frames import (
     LLMTextFrame,
 )
 from pipecat.metrics.metrics import LLMTokenUsage
+from pipecat.processors.aggregators.llm_context import LLMContext
 from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
 from pipecat.services.llm_service import FunctionCallFromLLM
 from pipecat.services.openai.llm import OpenAILLMService
@@ -99,7 +100,9 @@ class SambaNovaLLMService(OpenAILLMService):  # type: ignore
         return params
 
     @traced_llm  # type: ignore
-    async def _process_context(self, context: OpenAILLMContext) -> AsyncStream[ChatCompletionChunk]:
+    async def _process_context(
+        self, context: OpenAILLMContext | LLMContext
+    ) -> AsyncStream[ChatCompletionChunk]:
         """Process OpenAI LLM context and stream chat completion chunks.
 
         This method handles the streaming response from SambaNova API, including
@@ -122,9 +125,11 @@ class SambaNovaLLMService(OpenAILLMService):  # type: ignore
 
         await self.start_ttfb_metrics()
 
-        chunk_stream: AsyncStream[
-            ChatCompletionChunk
-        ] = await self._stream_chat_completions_specific_context(context)
+        chunk_stream = await (
+            self._stream_chat_completions_specific_context(context)
+            if isinstance(context, OpenAILLMContext)
+            else self._stream_chat_completions_universal_context(context)
+        )
 
         async for chunk in chunk_stream:
             if chunk.usage:
@@ -210,12 +215,3 @@ class SambaNovaLLMService(OpenAILLMService):  # type: ignore
                 )
 
             await self.run_function_calls(function_calls)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as SambaNovaLLMService does not yet support universal LLMContext.
-        """
-        return False

From 6838bc1e51eec72c7c4285c21dbdf18fb975159a Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:52:44 -0400
Subject: [PATCH 20/23] Add a few missing `LLMContext` type hints

---
 src/pipecat/services/grok/llm.py       | 3 ++-
 src/pipecat/services/nim/llm.py        | 3 ++-
 src/pipecat/services/perplexity/llm.py | 3 ++-
 3 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/src/pipecat/services/grok/llm.py b/src/pipecat/services/grok/llm.py
index 2a9704008..684be2d39 100644
--- a/src/pipecat/services/grok/llm.py
+++ b/src/pipecat/services/grok/llm.py
@@ -16,6 +16,7 @@ from dataclasses import dataclass
 from loguru import logger
 
 from pipecat.metrics.metrics import LLMTokenUsage
+from pipecat.processors.aggregators.llm_context import LLMContext
 from pipecat.processors.aggregators.llm_response import (
     LLMAssistantAggregatorParams,
     LLMUserAggregatorParams,
@@ -107,7 +108,7 @@ class GrokLLMService(OpenAILLMService):
         logger.debug(f"Creating Grok client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
 
-    async def _process_context(self, context: OpenAILLMContext):
+    async def _process_context(self, context: OpenAILLMContext | LLMContext):
         """Process a context through the LLM and accumulate token usage metrics.
 
         This method overrides the parent class implementation to handle Grok's
diff --git a/src/pipecat/services/nim/llm.py b/src/pipecat/services/nim/llm.py
index 052b94274..07e970521 100644
--- a/src/pipecat/services/nim/llm.py
+++ b/src/pipecat/services/nim/llm.py
@@ -11,6 +11,7 @@ Microservice) API while maintaining compatibility with the OpenAI-style interfac
 """
 
 from pipecat.metrics.metrics import LLMTokenUsage
+from pipecat.processors.aggregators.llm_context import LLMContext
 from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
 from pipecat.services.openai.llm import OpenAILLMService
 
@@ -47,7 +48,7 @@ class NimLLMService(OpenAILLMService):
         self._has_reported_prompt_tokens = False
         self._is_processing = False
 
-    async def _process_context(self, context: OpenAILLMContext):
+    async def _process_context(self, context: OpenAILLMContext | LLMContext):
         """Process a context through the LLM and accumulate token usage metrics.
 
         This method overrides the parent class implementation to handle NVIDIA's
diff --git a/src/pipecat/services/perplexity/llm.py b/src/pipecat/services/perplexity/llm.py
index bc4b220ac..814699070 100644
--- a/src/pipecat/services/perplexity/llm.py
+++ b/src/pipecat/services/perplexity/llm.py
@@ -15,6 +15,7 @@ from openai import NOT_GIVEN
 
 from pipecat.adapters.services.open_ai_adapter import OpenAILLMInvocationParams
 from pipecat.metrics.metrics import LLMTokenUsage
+from pipecat.processors.aggregators.llm_context import LLMContext
 from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
 from pipecat.services.openai.llm import OpenAILLMService
 
@@ -84,7 +85,7 @@ class PerplexityLLMService(OpenAILLMService):
 
         return params
 
-    async def _process_context(self, context: OpenAILLMContext):
+    async def _process_context(self, context: OpenAILLMContext | LLMContext):
         """Process a context through the LLM and accumulate token usage metrics.
 
         This method overrides the parent class implementation to handle

From 7cb372ebb9c5862bc6937957673b6a078320f92c Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 15:56:44 -0400
Subject: [PATCH 21/23] Add support for universal `LLMContext` to Together.ai
 LLM service

---
 src/pipecat/services/together/llm.py | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/src/pipecat/services/together/llm.py b/src/pipecat/services/together/llm.py
index 2a004f1c9..7a22c885a 100644
--- a/src/pipecat/services/together/llm.py
+++ b/src/pipecat/services/together/llm.py
@@ -49,12 +49,3 @@ class TogetherLLMService(OpenAILLMService):
         """
         logger.debug(f"Creating Together.ai client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
-
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as TogetherLLMService does not yet support universal LLMContext.
-        """
-        return False

From 9b216116f1707c638b0627933903c4ffc31821b8 Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 16:00:48 -0400
Subject: [PATCH 22/23] Remove `supports_universal_context` gate from
 OpenAI-library-based LLM services. It's no longer needed, as they all now
 support universal `LLMContext`.

---
 src/pipecat/services/google/llm_openai.py |  9 ---------
 src/pipecat/services/openai/base_llm.py   | 19 +------------------
 src/pipecat/services/openai/llm.py        |  9 ---------
 3 files changed, 1 insertion(+), 36 deletions(-)

diff --git a/src/pipecat/services/google/llm_openai.py b/src/pipecat/services/google/llm_openai.py
index 83d90e4ac..81d124cb7 100644
--- a/src/pipecat/services/google/llm_openai.py
+++ b/src/pipecat/services/google/llm_openai.py
@@ -76,15 +76,6 @@ class GoogleLLMOpenAIBetaService(OpenAILLMService):
 
         super().__init__(api_key=api_key, base_url=base_url, model=model, **kwargs)
 
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            False, as GoogleLLMOpenAIBetaService does not yet support universal LLMContext.
-        """
-        return False
-
     async def _process_context(self, context: OpenAILLMContext):
         functions_list = []
         arguments_list = []
diff --git a/src/pipecat/services/openai/base_llm.py b/src/pipecat/services/openai/base_llm.py
index 42eeb8981..682c0f227 100644
--- a/src/pipecat/services/openai/base_llm.py
+++ b/src/pipecat/services/openai/base_llm.py
@@ -419,18 +419,6 @@ class BaseOpenAILLMService(LLMService):
 
             await self.run_function_calls(function_calls)
 
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            Whether service supports universal LLMContext.
-        """
-        # Return True in subclasses that support universal LLMContext
-        # This property lets us gradually roll out support for universal
-        # LLMContext to OpenAI-like services in a controlled manner.
-        return False
-
     async def process_frame(self, frame: Frame, direction: FrameDirection):
         """Process frames for LLM completion requests.
 
@@ -450,12 +438,7 @@ class BaseOpenAILLMService(LLMService):
             context = frame.context
         elif isinstance(frame, LLMContextFrame):
             # Handle universal (LLM-agnostic) LLM context frames
-            if self.supports_universal_context:
-                context = frame.context
-            else:
-                raise NotImplementedError(
-                    f"Universal LLMContext is not yet supported for {self.__class__.__name__}."
-                )
+            context = frame.context
         elif isinstance(frame, LLMMessagesFrame):
             # NOTE: LLMMessagesFrame is deprecated, so we don't support the newer universal
             # LLMContext with it
diff --git a/src/pipecat/services/openai/llm.py b/src/pipecat/services/openai/llm.py
index 9f5d896b6..7919dd159 100644
--- a/src/pipecat/services/openai/llm.py
+++ b/src/pipecat/services/openai/llm.py
@@ -107,15 +107,6 @@ class OpenAILLMService(BaseOpenAILLMService):
         assistant = OpenAIAssistantContextAggregator(context, params=assistant_params)
         return OpenAIContextAggregatorPair(_user=user, _assistant=assistant)
 
-    @property
-    def supports_universal_context(self) -> bool:
-        """Check if this service supports universal LLMContext.
-
-        Returns:
-            True, as OpenAI service supports universal LLMContext.
-        """
-        return True
-
 
 class OpenAIUserContextAggregator(LLMUserContextAggregator):
     """OpenAI-specific user context aggregator.

From def1cf1548c879d35a997aaaa9856d155cc6946b Mon Sep 17 00:00:00 2001
From: Paul Kompfner <paul@daily.co>
Date: Tue, 2 Sep 2025 16:04:41 -0400
Subject: [PATCH 23/23] Update CHANGELOG with entry about expanded support
 among LLM services for new universal `LLMContext`

---
 CHANGELOG.md | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index af3969560..fe68d5f3c 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -9,6 +9,28 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Expanded support for universal `LLMContext` to more LLM services. Using the
+  universal `LLMContext` and associated `LLMContextAggregatorPair` is a
+  pre-requisite for using `LLMSwitcher` to switch between LLMs at runtime.
+  Here are the newly-supported services:
+
+  - Azure
+  - Cerebras
+  - Deepseek
+  - Fireworks AI
+  - Google Vertex AI
+  - Grok
+  - Groq
+  - Mistral
+  - NVIDIA NIM
+  - Ollama
+  - OpenPipe
+  - OpenRouter
+  - Perplexity
+  - Qwen
+  - SambaNova
+  - Together.ai
+
 - Added support for WhatsApp User-initiated Calls.
 
 - Added new audio filter `AICFilter`, speech enhancement for improving VAD/STT