unique_toolkit 1.55.0__tar.gz → 1.56.0__tar.gz

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: unique-toolkit
-Version: 1.55.0
+Version: 1.56.0
 Summary: 
 Author: Cedric Klinkert, Martin Fadler, Sadique Sheik, Fabian Schläpfer, Pascal Hauri, Ejiroghene Oki-Peter, Rami Azouz, Ricardo Visini, Sanzio Monti
 Author-email: Cedric Klinkert <cedric.klinkert@unique.ch>, Martin Fadler <martin.fadler@unique.ch>, Sadique Sheik <sadique@unique.ch>, Fabian Schläpfer <fabian@unique.ch>, Pascal Hauri <pascal@unique.ch>, Ejiroghene Oki-Peter <oki-peter.ejiroghene@unique.ai>, Rami Azouz <rami.ext@unique.ch>, Ricardo Visini <ricardo.visini@unique.ch>, Sanzio Monti <sanzio.monti@unique.ai>

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "unique_toolkit"
-version = "1.55.0"
+version = "1.56.0"
 description = ""
 readme = "README.md"
 requires-python = ">=3.12"

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/unique_toolkit/agentic/evaluation/evaluation_manager.py

@@ -1,4 +1,5 @@
 import asyncio
+import time
 from abc import ABC
 from logging import Logger
 
@@ -86,6 +87,7 @@ class EvaluationManager:
         self._chat_service = chat_service
         self._evaluations: dict[EvaluationMetricName, Evaluation] = {}
         self._evaluation_passed: bool = True
+        self._execution_times: dict[str, float] = {}
 
     def add_evaluation(self, evaluation: Evaluation):
         self._evaluations[evaluation.get_name()] = evaluation
@@ -99,6 +101,8 @@ class EvaluationManager:
         loop_response: LanguageModelStreamResponse,
         assistant_message_id: str,
     ) -> list[EvaluationMetricResult]:
+        self._execution_times = {}
+
         task_executor = SafeTaskExecutor(
             logger=self._logger,
         )
@@ -134,6 +138,9 @@ class EvaluationManager:
 
         return evaluation_results_unpacked
 
+    def get_execution_times(self) -> dict[str, float]:
+        return self._execution_times.copy()
+
     async def execute_evaluation_call(
         self,
         evaluation_name: EvaluationMetricName,
@@ -145,13 +152,16 @@ class EvaluationManager:
         evaluation_instance = self.get_evaluation_by_name(evaluation_name)
 
         if evaluation_instance:
-            # Execute the evaluation
+            start = time.perf_counter()
             await self._create_assistant_message(
                 evaluation_instance, assistant_message_id
             )
             evaluation_metric_result: EvaluationMetricResult = (
                 await evaluation_instance.run(loop_response)
             )
+            self._execution_times[str(evaluation_name)] = round(
+                time.perf_counter() - start, 3
+            )
             return evaluation_metric_result
 
         return EvaluationMetricResult(

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/unique_toolkit/agentic/postprocessor/postprocessor_manager.py

@@ -1,4 +1,5 @@
 import asyncio
+import time
 from abc import ABC
 from logging import Logger
 
@@ -84,6 +85,7 @@ class PostprocessorManager:
         self._logger = logger
         self._chat_service = chat_service
         self._postprocessors: list[Postprocessor | ResponsesApiPostprocessor] = []
+        self._execution_times: dict[str, float] = {}
 
         # Allow to add postprocessors that should be run before or after the others.
         self._first_postprocessor: Postprocessor | ResponsesApiPostprocessor | None = (
@@ -123,6 +125,8 @@ class PostprocessorManager:
         self,
         loop_response: LanguageModelStreamResponse,
     ) -> None:
+        self._execution_times = {}
+
         task_executor = SafeTaskExecutor(
             logger=self._logger,
         )
@@ -158,12 +162,19 @@ class PostprocessorManager:
                 references=loop_response.message.references,
             )
 
+    def get_execution_times(self) -> dict[str, float]:
+        return self._execution_times.copy()
+
     async def execute_postprocessors(
         self,
         loop_response: LanguageModelStreamResponse,
         postprocessor_instance: Postprocessor | ResponsesApiPostprocessor,
     ) -> None:
+        start = time.perf_counter()
         await postprocessor_instance.run(loop_response)  # type: ignore
+        self._execution_times[postprocessor_instance.name] = round(
+            time.perf_counter() - start, 3
+        )
 
     async def remove_from_text(
         self,

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/unique_toolkit/agentic/tools/tool_manager.py

@@ -1,4 +1,5 @@
 import asyncio
+import time
 from logging import Logger, getLogger
 from typing import Generic, Literal, TypeVar, overload
 
@@ -288,10 +289,16 @@ class _ToolManager(Generic[_ApiMode]):
             raise ValueError(f"Tool {tool_call.name} cannot be run")
 
         if tool_instance:
-            # Execute the tool
+            tool_start = time.perf_counter()
             tool_response: ToolCallResponse = await tool_instance.run(
                 tool_call=tool_call
             )
+            tool_execution_time = round(time.perf_counter() - tool_start, 3)
+
+            if tool_response.debug_info is None:
+                tool_response.debug_info = {}
+            tool_response.debug_info["execution_time_s"] = tool_execution_time
+
             evaluation_checks = tool_instance.evaluation_check_list()
             self._tool_evaluation_check_list.update(evaluation_checks)
 

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/unique_toolkit/language_model/infos.py

@@ -52,6 +52,8 @@ class LanguageModelName(StrEnum):
     AZURE_GPT_51_CODEX_MINI_2025_1113 = "AZURE_GPT_51_CODEX_MINI_2025_1113"
     AZURE_GPT_52_2025_1211 = "AZURE_GPT_52_2025_1211"
     AZURE_GPT_52_CHAT_2025_1211 = "AZURE_GPT_52_CHAT_2025_1211"
+    AZURE_GPT_54_2026_0305 = "AZURE_GPT_54_2026_0305"
+    AZURE_GPT_54_PRO_2026_0305 = "AZURE_GPT_54_PRO_2026_0305"
     AZURE_GPT_4o_2024_0513 = "AZURE_GPT_4o_2024_0513"
     AZURE_GPT_4o_2024_0806 = "AZURE_GPT_4o_2024_0806"
     AZURE_GPT_4o_2024_1120 = "AZURE_GPT_4o_2024_1120"
@@ -100,6 +102,8 @@ class LanguageModelName(StrEnum):
     LITELLM_OPENAI_GPT_51_THINKING = "litellm:openai-gpt-5-1-thinking"
     LITELLM_OPENAI_GPT_52 = "litellm:openai-gpt-5-2"
     LITELLM_OPENAI_GPT_52_THINKING = "litellm:openai-gpt-5-2-thinking"
+    LITELLM_OPENAI_GPT_54 = "litellm:openai-gpt-5-4"
+    LITELLM_OPENAI_GPT_54_THINKING = "litellm:openai-gpt-5-4-thinking"
     LITELLM_OPENAI_O1 = "litellm:openai-o1"
     LITELLM_OPENAI_O3 = "litellm:openai-o3"
     LITELLM_OPENAI_O3_DEEP_RESEARCH = "litellm:openai-o3-deep-research"
@@ -170,6 +174,8 @@ def get_encoder_name(model_name: LanguageModelName) -> EncoderName:
             | LMN.AZURE_GPT_51_CODEX_MINI_2025_1113
             | LMN.AZURE_GPT_52_2025_1211
             | LMN.AZURE_GPT_52_CHAT_2025_1211
+            | LMN.AZURE_GPT_54_2026_0305
+            | LMN.AZURE_GPT_54_PRO_2026_0305
             | LMN.AZURE_MODEL_ROUTER_2025_1118
             | LMN.LITELLM_OPENAI_GPT_5
             | LMN.LITELLM_OPENAI_GPT_5_MINI
@@ -180,6 +186,8 @@ def get_encoder_name(model_name: LanguageModelName) -> EncoderName:
             | LMN.LITELLM_OPENAI_GPT_51_THINKING
             | LMN.LITELLM_OPENAI_GPT_52
             | LMN.LITELLM_OPENAI_GPT_52_THINKING
+            | LMN.LITELLM_OPENAI_GPT_54
+            | LMN.LITELLM_OPENAI_GPT_54_THINKING
             | LMN.LITELLM_OPENAI_O1
             | LMN.LITELLM_OPENAI_O3
             | LMN.LITELLM_OPENAI_O3_DEEP_RESEARCH
@@ -591,10 +599,10 @@ class LanguageModelInfo(BaseModel):
                     info_cutoff_at=date(2024, 9, 30),
                     published_at=date(2025, 11, 13),
                     temperature_bounds=TemperatureBounds(
-                        min_temperature=1.0, max_temperature=1.0
+                        min_temperature=0.0, max_temperature=1.0
                     ),
                     default_options={
-                        "reasoning_effort": None,
+                        "reasoning_effort": "none",
                     },
                 )
             case LanguageModelName.AZURE_GPT_51_THINKING_2025_1113:
@@ -724,15 +732,15 @@ class LanguageModelInfo(BaseModel):
                         ModelCapabilities.VISION,
                     ],
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=400_000, token_limit_output=128_000
+                        token_limit_input=272_000, token_limit_output=128_000
                     ),
                     info_cutoff_at=date(2024, 9, 30),
                     published_at=date(2025, 12, 11),
                     temperature_bounds=TemperatureBounds(
-                        min_temperature=1.0, max_temperature=1.0
+                        min_temperature=0.0, max_temperature=1.0
                     ),
                     default_options={
-                        "reasoning_effort": None,
+                        "reasoning_effort": "none",
                     },
                 )
             case LanguageModelName.AZURE_GPT_52_CHAT_2025_1211:
@@ -761,6 +769,61 @@ class LanguageModelInfo(BaseModel):
                         "reasoning_effort": "medium",
                     },
                 )
+            case LanguageModelName.AZURE_GPT_54_2026_0305:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="2026-03-05",
+                    encoder_name=EncoderName.O200K_BASE,
+                    capabilities=[
+                        ModelCapabilities.CHAT_COMPLETIONS_API,
+                        ModelCapabilities.FUNCTION_CALLING,
+                        ModelCapabilities.PARALLEL_FUNCTION_CALLING,
+                        ModelCapabilities.REASONING,
+                        ModelCapabilities.RESPONSES_API,
+                        ModelCapabilities.STREAMING,
+                        ModelCapabilities.STRUCTURED_OUTPUT,
+                        ModelCapabilities.VISION,
+                    ],
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=922_000, token_limit_output=128_000
+                    ),
+                    info_cutoff_at=date(2024, 9, 30),
+                    published_at=date(2026, 3, 5),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=0.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "none",
+                    },
+                )
+            case LanguageModelName.AZURE_GPT_54_PRO_2026_0305:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.AZURE,
+                    version="2026-03-05",
+                    encoder_name=EncoderName.O200K_BASE,
+                    capabilities=[
+                        ModelCapabilities.FUNCTION_CALLING,
+                        ModelCapabilities.PARALLEL_FUNCTION_CALLING,
+                        ModelCapabilities.REASONING,
+                        ModelCapabilities.RESPONSES_API,
+                        ModelCapabilities.STREAMING,
+                        ModelCapabilities.STRUCTURED_OUTPUT,
+                        ModelCapabilities.VISION,
+                    ],
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=922_000, token_limit_output=128_000
+                    ),
+                    info_cutoff_at=date(2024, 9, 30),
+                    published_at=date(2026, 3, 5),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "medium",
+                    },
+                )
             case LanguageModelName.AZURE_GPT_4_TURBO_2024_0409:
                 return cls(
                     name=model_name,
@@ -1681,10 +1744,10 @@ class LanguageModelInfo(BaseModel):
                     info_cutoff_at=date(2024, 9, 30),
                     published_at=date(2025, 11, 13),
                     temperature_bounds=TemperatureBounds(
-                        min_temperature=1.0, max_temperature=1.0
+                        min_temperature=0.0, max_temperature=1.0
                     ),
                     default_options={
-                        "reasoning_effort": None,
+                        "reasoning_effort": "none",
                     },
                 )
             case LanguageModelName.LITELLM_OPENAI_GPT_51_THINKING:
@@ -1732,15 +1795,15 @@ class LanguageModelInfo(BaseModel):
                         ModelCapabilities.VISION,
                     ],
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=400_000, token_limit_output=128_000
+                        token_limit_input=272_000, token_limit_output=128_000
                     ),
                     info_cutoff_at=date(2024, 9, 30),
                     published_at=date(2025, 12, 11),
                     temperature_bounds=TemperatureBounds(
-                        min_temperature=1.0, max_temperature=1.0
+                        min_temperature=0.0, max_temperature=1.0
                     ),
                     default_options={
-                        "reasoning_effort": None,
+                        "reasoning_effort": "none",
                     },
                 )
             case LanguageModelName.LITELLM_OPENAI_GPT_52_THINKING:
@@ -1760,7 +1823,7 @@ class LanguageModelInfo(BaseModel):
                         ModelCapabilities.VISION,
                     ],
                     token_limits=LanguageModelTokenLimits(
-                        token_limit_input=400_000, token_limit_output=128_000
+                        token_limit_input=272_000, token_limit_output=128_000
                     ),
                     info_cutoff_at=date(2024, 9, 30),
                     published_at=date(2025, 12, 11),
@@ -1771,6 +1834,62 @@ class LanguageModelInfo(BaseModel):
                         "reasoning_effort": "medium",
                     },
                 )
+            case LanguageModelName.LITELLM_OPENAI_GPT_54:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.LITELLM,
+                    version="2026-03-05",
+                    encoder_name=EncoderName.O200K_BASE,
+                    capabilities=[
+                        ModelCapabilities.CHAT_COMPLETIONS_API,
+                        ModelCapabilities.FUNCTION_CALLING,
+                        ModelCapabilities.PARALLEL_FUNCTION_CALLING,
+                        ModelCapabilities.REASONING,
+                        ModelCapabilities.RESPONSES_API,
+                        ModelCapabilities.STREAMING,
+                        ModelCapabilities.STRUCTURED_OUTPUT,
+                        ModelCapabilities.VISION,
+                    ],
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=922_000, token_limit_output=128_000
+                    ),
+                    info_cutoff_at=date(2024, 9, 30),
+                    published_at=date(2026, 3, 5),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=0.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "none",
+                    },
+                )
+            case LanguageModelName.LITELLM_OPENAI_GPT_54_THINKING:
+                return cls(
+                    name=model_name,
+                    provider=LanguageModelProvider.LITELLM,
+                    version="2026-03-05",
+                    encoder_name=EncoderName.O200K_BASE,
+                    capabilities=[
+                        ModelCapabilities.CHAT_COMPLETIONS_API,
+                        ModelCapabilities.FUNCTION_CALLING,
+                        ModelCapabilities.PARALLEL_FUNCTION_CALLING,
+                        ModelCapabilities.REASONING,
+                        ModelCapabilities.RESPONSES_API,
+                        ModelCapabilities.STREAMING,
+                        ModelCapabilities.STRUCTURED_OUTPUT,
+                        ModelCapabilities.VISION,
+                    ],
+                    token_limits=LanguageModelTokenLimits(
+                        token_limit_input=922_000, token_limit_output=128_000
+                    ),
+                    info_cutoff_at=date(2024, 9, 30),
+                    published_at=date(2026, 3, 5),
+                    temperature_bounds=TemperatureBounds(
+                        min_temperature=1.0, max_temperature=1.0
+                    ),
+                    default_options={
+                        "reasoning_effort": "medium",
+                    },
+                )
             case LanguageModelName.LITELLM_OPENAI_O1:
                 return cls(
                     name=model_name,

{unique_toolkit-1.55.0 → unique_toolkit-1.56.0}/unique_toolkit/services/chat_service.py

@@ -140,6 +140,50 @@ class ChatService(ChatServiceDeprecated):
 
         return self._elicitation_service
 
+    def get_debug_info(self) -> dict[str, Any]:
+        """Retrieves the debug information from the current user message.
+
+        Returns:
+            dict[str, Any]: The debug information stored on the user message,
+                or an empty dict if retrieval fails or no debug info exists.
+
+        """
+        try:
+            raw_msg = unique_sdk.Message.retrieve(
+                user_id=self._user_id,
+                company_id=self._company_id,
+                id=self._user_message_id,
+                chatId=self._chat_id,
+            )
+            return getattr(raw_msg, "debugInfo", {}) or {}
+        except Exception:
+            logging.getLogger(__name__).warning(
+                "Failed to retrieve debug info from user message", exc_info=True
+            )
+            return {}
+
+    async def get_debug_info_async(self) -> dict[str, Any]:
+        """Retrieves the debug information from the current user message asynchronously.
+
+        Returns:
+            dict[str, Any]: The debug information stored on the user message,
+                or an empty dict if retrieval fails or no debug info exists.
+
+        """
+        try:
+            raw_msg = await unique_sdk.Message.retrieve_async(
+                user_id=self._user_id,
+                company_id=self._company_id,
+                id=self._user_message_id,
+                chatId=self._chat_id,
+            )
+            return getattr(raw_msg, "debugInfo", {}) or {}
+        except Exception:
+            logging.getLogger(__name__).warning(
+                "Failed to retrieve debug info from user message", exc_info=True
+            )
+            return {}
+
     async def update_debug_info_async(self, debug_info: dict):
         """Updates the debug information for the chat session.