arize-phoenix 11.10.0__py3-none-any.whl → 11.11.0__py3-none-any.whl

{arize_phoenix-11.10.0.dist-info → arize_phoenix-11.11.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arize-phoenix
-Version: 11.10.0
+Version: 11.11.0
 Summary: AI Observability and Evaluation
 Project-URL: Documentation, https://arize.com/docs/phoenix/
 Project-URL: Issues, https://github.com/Arize-ai/phoenix/issues

{arize_phoenix-11.10.0.dist-info → arize_phoenix-11.11.0.dist-info}/RECORD

@@ -6,7 +6,7 @@ phoenix/exceptions.py,sha256=n2L2KKuecrdflB9MsCdAYCiSEvGJptIsfRkXMoJle7A,169
 phoenix/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 phoenix/services.py,sha256=ngkyKGVatX3cO2WJdo2hKdaVKP-xJCMvqthvga6kJss,5196
 phoenix/settings.py,sha256=2kHfT3BNOVd4dAO1bq-syEQbHSG8oX2-7NhOwK2QREk,896
-phoenix/version.py,sha256=QWduDFvGZEp2yqC4pdaQmCm9ifKoDwiu-l0lM7_65A0,24
+phoenix/version.py,sha256=0NvGC949F8-jU5CThwqqkP5g0LPLHwkhtLR2cEkRUTU,24
 phoenix/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/core/embedding_dimension.py,sha256=zKGbcvwOXgLf-yrJBpQyKtd-LEOPRKHnUToyAU8Owis,87
 phoenix/core/model.py,sha256=qBFraOtmwCCnWJltKNP18DDG0mULXigytlFsa6YOz6k,4837
@@ -112,7 +112,7 @@ phoenix/server/api/auth.py,sha256=AyYhnZIbY9ALVjg2K6aC2UXSa3Pva5GVDBXyaZ3nD3o,27
 phoenix/server/api/context.py,sha256=mqsq_8Ru50e-PxKWNTzh9zptb1PFjYFUf58uW59UYL0,8996
 phoenix/server/api/exceptions.py,sha256=E2W0x63CBzc0CoQPptrLr9nZxPF9zIP8MCJ3RuJMddw,1322
 phoenix/server/api/interceptor.py,sha256=ykDnoC_apUd-llVli3m1CW18kNSIgjz2qZ6m5JmPDu8,1294
-phoenix/server/api/queries.py,sha256=fvbdyhJ57I6DPrNEBkWwhw8BRF7DyIZ3LtYwXwI7yVw,45651
+phoenix/server/api/queries.py,sha256=EVNKWanVbjcPIfG8zW8_p3bBkLxmNuBbS87NUdW68z4,46194
 phoenix/server/api/schema.py,sha256=fcs36xQwFF_Qe41_5cWR8wYpDvOrnbcyTeo5WNMbDsA,1702
 phoenix/server/api/subscriptions.py,sha256=ZOGNsLVr5TNjCWgbzO7Eq6Ls_NRdJH9AxC0cW_v0vhM,25332
 phoenix/server/api/utils.py,sha256=quCBRcusc6PUq9tJq7M8PgwFZp7nXgVAxtbw8feribY,833
@@ -176,8 +176,8 @@ phoenix/server/api/helpers/__init__.py,sha256=m2-xaSPqUiSs91k62JaRDjFNfl-1byxBfY
 phoenix/server/api/helpers/annotations.py,sha256=9gMXKpMTfWEChoSCnvdWYuyB0hlSnNOp-qUdar9Vono,262
 phoenix/server/api/helpers/dataset_helpers.py,sha256=3bdGBoUzqrtg-sr5p2wpQLOU6dhg_3TKFHNeJj8p0TU,9155
 phoenix/server/api/helpers/experiment_run_filters.py,sha256=DOnVwrmn39eAkk2mwuZP8kIcAnR5jrOgllEwWSjsw94,29893
-phoenix/server/api/helpers/playground_clients.py,sha256=vcLc8Ps8ZaZCUJnsN7retFbUR0LjF8nKTtN_OjBfS3E,71172
-phoenix/server/api/helpers/playground_registry.py,sha256=CPLMziFB2wmr-dfbx7VbzO2f8YIG_k5RftzvGXYGQ1w,2570
+phoenix/server/api/helpers/playground_clients.py,sha256=Fq4DNVIdnCiiVt0bh5mrZ7dJb2oOQcLjTttfq0Wcuv0,73589
+phoenix/server/api/helpers/playground_registry.py,sha256=n0v4-KnvZJxeaEwOla5qBbnOQjSWznKmMhZnh9ziJt0,2584
 phoenix/server/api/helpers/playground_spans.py,sha256=QpXwPl_fFNwm_iA1A77XApUyXMl1aDmonw8aXuNZ_4k,17132
 phoenix/server/api/helpers/prompts/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/server/api/helpers/prompts/models.py,sha256=nlPtLZaGcHfWNRR0iNRaBUv8eoKOnoGqRm6zadrTt0I,23547
@@ -218,6 +218,7 @@ phoenix/server/api/input_types/PerformanceMetricInput.py,sha256=4SG5AymxV2dMeCrK
 phoenix/server/api/input_types/ProjectFilter.py,sha256=w3IimMXcNmMkrQ6h89-Yx6uh21r2wj1jci1nPZEDEhQ,260
 phoenix/server/api/input_types/ProjectSessionSort.py,sha256=KZzEtOMpcxtP11brL4fXUAY_RLAK-Ul4luAWwVKnQgY,1043
 phoenix/server/api/input_types/ProjectSort.py,sha256=ZTT-InFB6NvInDwLuivyHe9PxR5nsmQ8aXCHAPjZm9k,329
+phoenix/server/api/input_types/PromptFilter.py,sha256=f2F7fDlYRsNJp_rKkmvpgUt9rfgr_e-dyZxuHX8YfkU,256
 phoenix/server/api/input_types/PromptTemplateOptions.py,sha256=8ZJdH1F9fExcdH9dF8SJ29WycCvtEpK-Z6dZwFO7KgQ,232
 phoenix/server/api/input_types/PromptVersionInput.py,sha256=n6zBeSkK8ZFRHTjtVx4BK--azZIxXeYETa6Cufcet2I,3743
 phoenix/server/api/input_types/SpanAnnotationFilter.py,sha256=-djfIXYCxV6sV3GPOZQUV0SPfiWDhRlTORfeQ7tCBgQ,2671
@@ -308,7 +309,7 @@ phoenix/server/api/types/ExperimentRun.py,sha256=_fcwDLuURV0yviOlkjWAgJJwcCPdz-x
 phoenix/server/api/types/ExperimentRunAnnotation.py,sha256=YGw5zIbjRXUK3zH475DnEeg4SDNGOmdxtuUVkzGw1E8,1734
 phoenix/server/api/types/ExportedFile.py,sha256=e3GTn7B5LgsTbqiwjhMCQH7VsiqXitrBO4aCMS1lHsg,163
 phoenix/server/api/types/Functionality.py,sha256=zDDl2bANIqjwfooSOHg-VQk6-wQy05mREwjV_-VbSIg,262
-phoenix/server/api/types/GenerativeModel.py,sha256=cAZdZ0dgOwrbJqIInP2Iirk3CYhQWK8dGwLvKfJeJlI,5814
+phoenix/server/api/types/GenerativeModel.py,sha256=HHpM2aznlmOWeZY77NzTYOcB9iIvWnfbjNUxIOJtvIQ,7688
 phoenix/server/api/types/GenerativeProvider.py,sha256=blXHIzZwe-xlu3-iF2dexvqnb4xxiD2XynS4Vw0iLx4,6750
 phoenix/server/api/types/Identifier.py,sha256=n3rxpoKNCwEvZu7QY8yr7g3AW2mU-U62BxFXYaiHLKk,306
 phoenix/server/api/types/InferenceModel.py,sha256=VWX7eKehFxXp4i_QW_dJAGzeOvh418VUt15AaT_Cv68,8090
@@ -320,7 +321,7 @@ phoenix/server/api/types/ModelInterface.py,sha256=Qe7H23wDb_Q2-HmeY2t0R5Jsn4aAfY
 phoenix/server/api/types/NumericRange.py,sha256=afEjgF97Go_OvmjMggbPBt-zGM8IONewAyEiKEHRds0,192
 phoenix/server/api/types/PerformanceMetric.py,sha256=KFkmJDqP43eDUtARQOUqR7NYcxvL6Vh2uisHWU6H3ko,387
 phoenix/server/api/types/PlaygroundModel.py,sha256=IqJFxsAAJMRyaFI9ryI3GQrpFOJ5Llf6kIutEO-tFvM,321
-phoenix/server/api/types/Project.py,sha256=FE93mSrWKv1rcCPECfp1hsCen2012SPp-1pCEVuq5ic,64039
+phoenix/server/api/types/Project.py,sha256=wL4jihXtyWjxAtSyRXBpz0DTH-vWZZBMhQFzBBFkl2A,69628
 phoenix/server/api/types/ProjectSession.py,sha256=uwqTsDTfSGz13AvP-cwS_mJR5JZ1lHqu10ungbl7g5s,6245
 phoenix/server/api/types/ProjectTraceRetentionPolicy.py,sha256=tYy2kgalPDyuaYZr0VUHjH0YpXaiF_QOzg5yfaV_c7c,3782
 phoenix/server/api/types/Prompt.py,sha256=ccP4eq1e38xbF0afclGWLOuDpBVpNbJ3AOSRClF8yFQ,4955
@@ -387,17 +388,17 @@ phoenix/server/static/apple-touch-icon-76x76.png,sha256=CT_xT12I0u2i0WU8JzBZBuOQ
 phoenix/server/static/apple-touch-icon.png,sha256=fOfpjqGpWYbJ0eAurKsyoZP1EAs6ZVooBJ_SGk2ZkDs,3801
 phoenix/server/static/favicon.ico,sha256=bY0vvCKRftemZfPShwZtE93DiiQdaYaozkPGwNFr6H8,34494
 phoenix/server/static/modernizr.js,sha256=mvK-XtkNqjOral-QvzoqsyOMECXIMu5BQwSVN_wcU9c,2564
-phoenix/server/static/.vite/manifest.json,sha256=67Jr1Pbvt4QUZP4O98Fe_9tPLJvVHSKqSJeYcfC2oVk,2165
-phoenix/server/static/assets/components-XAeml0-1.js,sha256=aCeHz5UUtnMLqWDeVRMmHJnjMHgwEDjpdRoz_tJ063s,620501
-phoenix/server/static/assets/index-D7EtHUpz.js,sha256=FBp-_XLUtdcJ3wLAdu0peRKAPTkU2Ceu_U2Fu-QRfDg,62601
-phoenix/server/static/assets/pages-CPfaxiKa.js,sha256=EyoqjFXTFOgpww-GDBYnH7KfsLfV84OI06WYG0xmUjo,1190498
+phoenix/server/static/.vite/manifest.json,sha256=FDZDQSSV67fIjs_K97eQRfjkSjyK76N2JPn2lzPBVPo,2165
+phoenix/server/static/assets/components-B7lK-RgC.js,sha256=u8Q0_GhFNtht2Js6HZflznlkqdT2FF3rUUFwq3pYZcU,630858
+phoenix/server/static/assets/index-CpePoyTU.js,sha256=LH2IdynxWg7bTImdqHOzUfNtTN9qd5XjFF0nINOdSlM,63028
+phoenix/server/static/assets/pages-CnVg_GUi.js,sha256=V9PbqoJu_CNMRy-4qCeLNGhMnGplEzsRDVHb4-ca9ZI,1198572
 phoenix/server/static/assets/vendor-CqDb5u4o.css,sha256=zIyFiNJKxMaQk8AvtLgt1rR01oO10d1MFndSDKH9Clw,5517
-phoenix/server/static/assets/vendor-DhvamIr8.js,sha256=hZdQcaVySUPWh4GXRhuL-PhUy2K4sJLs6gYxde6qB_I,2748936
-phoenix/server/static/assets/vendor-arizeai-4fVwwnrI.js,sha256=8eBZfGf2fjjNoWDW31mHXdxkqSlWItGj7eHhbnH92Qk,151750
-phoenix/server/static/assets/vendor-codemirror-DRfFHb57.js,sha256=v-q8mq6f-EbjtvgzLTQYF-SNKX4R6LOGJe5ucOMqx1w,781264
-phoenix/server/static/assets/vendor-recharts-w6bSawXG.js,sha256=uTuxE0vslP7_y-F1dOU5K2nXke7qe8JMdE2SnRGCVKM,231651
-phoenix/server/static/assets/vendor-shiki-CplrhwOk.js,sha256=ISu7sYmhh_FDTmnBN-icbipm6fa2InqOeQTo2JFr3LI,8980312
+phoenix/server/static/assets/vendor-arizeai-CXCKGfvH.js,sha256=TbpgJ-co-S2Op0E39YiNpafbzoe1OucxLp6g6-wbQNg,151750
+phoenix/server/static/assets/vendor-codemirror-DWr46-WB.js,sha256=DrigLnd9LH-JjswljzWQD1jU2OZ7MaCdwS4u5MtVUh0,553698
+phoenix/server/static/assets/vendor-recharts-0Yf6lanX.js,sha256=56vcRJWCY3T1Qr1H-tWgKVPhoG_tA2iwPm_KA9whHVY,231651
+phoenix/server/static/assets/vendor-shiki-Caei6iKO.js,sha256=oprmwuXapui4i3qdoe4wLxgn0Iotq2xDRG8RwzROPco,8980312
 phoenix/server/static/assets/vendor-three-C5WAXd5r.js,sha256=ELkg06u70N7h8oFmvqdoHyPuUf9VgGEWeT4LKFx4VWo,620975
+phoenix/server/static/assets/vendor-uWG2jYEi.js,sha256=Pm6mq_o91KvR7cdHB3-rW3E8GVE-H-N1Cp0x1XLDJic,2681983
 phoenix/server/templates/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/server/templates/index.html,sha256=3VMDmbxYwo3OoqiQyFojU6JaMLKr5k8rITacYS7HTbs,6922
 phoenix/session/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -437,9 +438,9 @@ phoenix/utilities/project.py,sha256=auVpARXkDb-JgeX5f2aStyFIkeKvGwN9l7qrFeJMVxI,
 phoenix/utilities/re.py,sha256=6YyUWIkv0zc2SigsxfOWIHzdpjKA_TZo2iqKq7zJKvw,2081
 phoenix/utilities/span_store.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/utilities/template_formatters.py,sha256=gh9PJD6WEGw7TEYXfSst1UR4pWWwmjxMLrDVQ_CkpkQ,2779
-arize_phoenix-11.10.0.dist-info/METADATA,sha256=-AwyriOi2Hypd_vYo7mrHEKmXIoypvFLTTmwn_MkdKA,30851
-arize_phoenix-11.10.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-arize_phoenix-11.10.0.dist-info/entry_points.txt,sha256=Pgpn8Upxx9P8z8joPXZWl2LlnAlGc3gcQoVchb06X1Q,94
-arize_phoenix-11.10.0.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
-arize_phoenix-11.10.0.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
-arize_phoenix-11.10.0.dist-info/RECORD,,
+arize_phoenix-11.11.0.dist-info/METADATA,sha256=EeNefvxStTYsByuT4E2kzd4DsmNHCrHNalV-nFxAV9A,30851
+arize_phoenix-11.11.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+arize_phoenix-11.11.0.dist-info/entry_points.txt,sha256=Pgpn8Upxx9P8z8joPXZWl2LlnAlGc3gcQoVchb06X1Q,94
+arize_phoenix-11.11.0.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
+arize_phoenix-11.11.0.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
+arize_phoenix-11.11.0.dist-info/RECORD,,

phoenix/server/api/helpers/playground_clients.py

@@ -20,7 +20,7 @@ from openinference.semconv.trace import (
 )
 from strawberry import UNSET
 from strawberry.scalars import JSON as JSONScalarType
-from typing_extensions import TypeAlias, assert_never
+from typing_extensions import TypeAlias, assert_never, override
 
 from phoenix.config import getenv
 from phoenix.evals.models.rate_limiters import (
@@ -437,9 +437,9 @@ class OpenAIBaseStreamingClient(PlaygroundStreamingClient):
         if role is ChatCompletionMessageRole.TOOL:
             if tool_call_id is None:
                 raise ValueError("tool_call_id is required for tool messages")
-        return ChatCompletionToolMessageParam(
-            {"content": content, "role": "tool", "tool_call_id": tool_call_id}
-        )
+            return ChatCompletionToolMessageParam(
+                {"content": content, "role": "tool", "tool_call_id": tool_call_id}
+            )
         assert_never(role)
 
     def to_openai_tool_call_param(
@@ -1140,27 +1140,28 @@ class OpenAIStreamingClient(OpenAIBaseStreamingClient):
         self._attributes[LLM_SYSTEM] = OpenInferenceLLMSystemValues.OPENAI.value
 
 
-@register_llm_client(
-    provider_key=GenerativeProviderKey.OPENAI,
-    model_names=[
-        "o1",
-        "o1-pro",
-        "o1-2024-12-17",
-        "o1-pro-2025-03-19",
-        "o1-mini",
-        "o1-mini-2024-09-12",
-        "o1-preview",
-        "o1-preview-2024-09-12",
-        "o3",
-        "o3-pro",
-        "o3-2025-04-16",
-        "o3-mini",
-        "o3-mini-2025-01-31",
-        "o4-mini",
-        "o4-mini-2025-04-16",
-    ],
-)
-class OpenAIReasoningStreamingClient(OpenAIStreamingClient):
+_OPENAI_REASONING_MODELS = [
+    "o1",
+    "o1-pro",
+    "o1-2024-12-17",
+    "o1-pro-2025-03-19",
+    "o1-mini",
+    "o1-mini-2024-09-12",
+    "o1-preview",
+    "o1-preview-2024-09-12",
+    "o3",
+    "o3-pro",
+    "o3-2025-04-16",
+    "o3-mini",
+    "o3-mini-2025-01-31",
+    "o4-mini",
+    "o4-mini-2025-04-16",
+]
+
+
+class OpenAIReasoningReasoningModelsMixin:
+    """Mixin class for OpenAI-style reasoning model clients (o1, o3 series)."""
+
     @classmethod
     def supported_invocation_parameters(cls) -> list[InvocationParameter]:
         return [
@@ -1191,6 +1192,16 @@ class OpenAIReasoningStreamingClient(OpenAIStreamingClient):
             ),
         ]
 
+
+@register_llm_client(
+    provider_key=GenerativeProviderKey.OPENAI,
+    model_names=_OPENAI_REASONING_MODELS,
+)
+class OpenAIReasoningNonStreamingClient(
+    OpenAIReasoningReasoningModelsMixin,
+    OpenAIStreamingClient,
+):
+    @override
     async def chat_completion_create(
         self,
         messages: list[
@@ -1283,46 +1294,11 @@ class OpenAIReasoningStreamingClient(OpenAIStreamingClient):
         if role is ChatCompletionMessageRole.TOOL:
             if tool_call_id is None:
                 raise ValueError("tool_call_id is required for tool messages")
-        return ChatCompletionToolMessageParam(
-            {"content": content, "role": "tool", "tool_call_id": tool_call_id}
-        )
+            return ChatCompletionToolMessageParam(
+                {"content": content, "role": "tool", "tool_call_id": tool_call_id}
+            )
         assert_never(role)
 
-    @staticmethod
-    def _llm_token_counts(usage: "CompletionUsage") -> Iterator[tuple[str, Any]]:
-        yield LLM_TOKEN_COUNT_PROMPT, usage.prompt_tokens
-        yield LLM_TOKEN_COUNT_COMPLETION, usage.completion_tokens
-        yield LLM_TOKEN_COUNT_TOTAL, usage.total_tokens
-
-        if hasattr(usage, "prompt_tokens_details") and usage.prompt_tokens_details is not None:
-            prompt_details = usage.prompt_tokens_details
-            if (
-                hasattr(prompt_details, "cached_tokens")
-                and prompt_details.cached_tokens is not None
-            ):
-                yield LLM_TOKEN_COUNT_PROMPT_DETAILS_CACHE_READ, prompt_details.cached_tokens
-            if hasattr(prompt_details, "audio_tokens") and prompt_details.audio_tokens is not None:
-                yield LLM_TOKEN_COUNT_PROMPT_DETAILS_AUDIO, prompt_details.audio_tokens
-
-        if (
-            hasattr(usage, "completion_tokens_details")
-            and usage.completion_tokens_details is not None
-        ):
-            completion_details = usage.completion_tokens_details
-            if (
-                hasattr(completion_details, "reasoning_tokens")
-                and completion_details.reasoning_tokens is not None
-            ):
-                yield (
-                    LLM_TOKEN_COUNT_COMPLETION_DETAILS_REASONING,
-                    completion_details.reasoning_tokens,
-                )
-            if (
-                hasattr(completion_details, "audio_tokens")
-                and completion_details.audio_tokens is not None
-            ):
-                yield LLM_TOKEN_COUNT_COMPLETION_DETAILS_AUDIO, completion_details.audio_tokens
-
 
 @register_llm_client(
     provider_key=GenerativeProviderKey.AZURE_OPENAI,
@@ -1376,6 +1352,113 @@ class AzureOpenAIStreamingClient(OpenAIBaseStreamingClient):
         self._attributes[LLM_SYSTEM] = OpenInferenceLLMSystemValues.OPENAI.value
 
 
+@register_llm_client(
+    provider_key=GenerativeProviderKey.AZURE_OPENAI,
+    model_names=_OPENAI_REASONING_MODELS,
+)
+class AzureOpenAIReasoningNonStreamingClient(
+    OpenAIReasoningReasoningModelsMixin,
+    AzureOpenAIStreamingClient,
+):
+    @override
+    async def chat_completion_create(
+        self,
+        messages: list[
+            tuple[ChatCompletionMessageRole, str, Optional[str], Optional[list[JSONScalarType]]]
+        ],
+        tools: list[JSONScalarType],
+        **invocation_parameters: Any,
+    ) -> AsyncIterator[ChatCompletionChunk]:
+        from openai import NOT_GIVEN
+
+        # Convert standard messages to OpenAI messages
+        openai_messages = []
+        for message in messages:
+            openai_message = self.to_openai_chat_completion_param(*message)
+            if openai_message is not None:
+                openai_messages.append(openai_message)
+
+        throttled_create = self.rate_limiter._alimit(self.client.chat.completions.create)
+        response = await throttled_create(
+            messages=openai_messages,
+            model=self.model_name,
+            stream=False,
+            tools=tools or NOT_GIVEN,
+            **invocation_parameters,
+        )
+
+        if response.usage is not None:
+            self._attributes.update(dict(self._llm_token_counts(response.usage)))
+
+        choice = response.choices[0]
+        if choice.message.content:
+            yield TextChunk(content=choice.message.content)
+
+        if choice.message.tool_calls:
+            for tool_call in choice.message.tool_calls:
+                yield ToolCallChunk(
+                    id=tool_call.id,
+                    function=FunctionCallChunk(
+                        name=tool_call.function.name,
+                        arguments=tool_call.function.arguments,
+                    ),
+                )
+
+    def to_openai_chat_completion_param(
+        self,
+        role: ChatCompletionMessageRole,
+        content: JSONScalarType,
+        tool_call_id: Optional[str] = None,
+        tool_calls: Optional[list[JSONScalarType]] = None,
+    ) -> Optional["ChatCompletionMessageParam"]:
+        from openai.types.chat import (
+            ChatCompletionAssistantMessageParam,
+            ChatCompletionDeveloperMessageParam,
+            ChatCompletionToolMessageParam,
+            ChatCompletionUserMessageParam,
+        )
+
+        if role is ChatCompletionMessageRole.USER:
+            return ChatCompletionUserMessageParam(
+                {
+                    "content": content,
+                    "role": "user",
+                }
+            )
+        if role is ChatCompletionMessageRole.SYSTEM:
+            return ChatCompletionDeveloperMessageParam(
+                {
+                    "content": content,
+                    "role": "developer",
+                }
+            )
+        if role is ChatCompletionMessageRole.AI:
+            if tool_calls is None:
+                return ChatCompletionAssistantMessageParam(
+                    {
+                        "content": content,
+                        "role": "assistant",
+                    }
+                )
+            else:
+                return ChatCompletionAssistantMessageParam(
+                    {
+                        "content": content,
+                        "role": "assistant",
+                        "tool_calls": [
+                            self.to_openai_tool_call_param(tool_call) for tool_call in tool_calls
+                        ],
+                    }
+                )
+        if role is ChatCompletionMessageRole.TOOL:
+            if tool_call_id is None:
+                raise ValueError("tool_call_id is required for tool messages")
+            return ChatCompletionToolMessageParam(
+                {"content": content, "role": "tool", "tool_call_id": tool_call_id}
+            )
+        assert_never(role)
+
+
 @register_llm_client(
     provider_key=GenerativeProviderKey.ANTHROPIC,
     model_names=[

phoenix/server/api/helpers/playground_registry.py

@@ -1,4 +1,4 @@
-from typing import TYPE_CHECKING, Any, Callable, Optional, Union
+from typing import TYPE_CHECKING, Any, Callable, Optional, Sequence, Union
 
 from phoenix.server.api.types.GenerativeProvider import GenerativeProviderKey
 
@@ -59,7 +59,7 @@ PLAYGROUND_CLIENT_REGISTRY: PlaygroundClientRegistry = PlaygroundClientRegistry(
 
 def register_llm_client(
     provider_key: GenerativeProviderKey,
-    model_names: list[ModelName],
+    model_names: Sequence[ModelName],
 ) -> Callable[[type["PlaygroundStreamingClient"]], type["PlaygroundStreamingClient"]]:
     def decorator(cls: type["PlaygroundStreamingClient"]) -> type["PlaygroundStreamingClient"]:
         provider_registry = PLAYGROUND_CLIENT_REGISTRY._registry.setdefault(provider_key, {})

phoenix/server/api/input_types/PromptFilter.py

@@ -0,0 +1,14 @@
+from enum import Enum
+
+import strawberry
+
+
+@strawberry.enum
+class PromptFilterColumn(Enum):
+    name = "name"
+
+
+@strawberry.input(description="The filter key and value for prompt connections")
+class PromptFilter:
+    col: PromptFilterColumn
+    value: str

phoenix/server/api/queries.py

@@ -1,12 +1,13 @@
 import re
 from collections import defaultdict
 from datetime import datetime
-from typing import Iterable, Iterator, Optional, Union, cast
+from typing import Iterable, Iterator, Optional, Union
+from typing import cast as type_cast
 
 import numpy as np
 import numpy.typing as npt
 import strawberry
-from sqlalchemy import and_, distinct, func, select, text
+from sqlalchemy import String, and_, cast, distinct, func, select, text
 from sqlalchemy.orm import joinedload
 from starlette.authentication import UnauthenticatedUser
 from strawberry import ID, UNSET
@@ -41,6 +42,7 @@ from phoenix.server.api.input_types.DatasetSort import DatasetSort
 from phoenix.server.api.input_types.InvocationParameters import InvocationParameter
 from phoenix.server.api.input_types.ProjectFilter import ProjectFilter
 from phoenix.server.api.input_types.ProjectSort import ProjectColumn, ProjectSort
+from phoenix.server.api.input_types.PromptFilter import PromptFilter
 from phoenix.server.api.types.AnnotationConfig import AnnotationConfig, to_gql_annotation_config
 from phoenix.server.api.types.Cluster import Cluster, to_gql_clusters
 from phoenix.server.api.types.Dataset import Dataset, to_gql_dataset
@@ -728,6 +730,7 @@ class Query:
         last: Optional[int] = UNSET,
         after: Optional[CursorString] = UNSET,
         before: Optional[CursorString] = UNSET,
+        filter: Optional[PromptFilter] = UNSET,
     ) -> Connection[Prompt]:
         args = ConnectionArgs(
             first=first,
@@ -736,6 +739,14 @@ class Query:
             before=before if isinstance(before, CursorString) else None,
         )
         stmt = select(models.Prompt)
+        if filter:
+            column = getattr(models.Prompt, filter.col.value)
+            # Cast Identifier columns to String for ilike operations
+            if filter.col.value == "name":
+                column = cast(column, String)
+            stmt = stmt.where(column.ilike(f"%{filter.value}%")).order_by(
+                models.Prompt.updated_at.desc()
+            )
         async with info.context.db() as session:
             orm_prompts = await session.stream_scalars(stmt)
             data = [to_gql_prompt_from_orm(orm_prompt) async for orm_prompt in orm_prompts]
@@ -994,7 +1005,7 @@ class Query:
             """).bindparams(nspname=nspname)
             try:
                 async with info.context.db() as session:
-                    stats = cast(Iterable[tuple[str, int]], await session.execute(stmt))
+                    stats = type_cast(Iterable[tuple[str, int]], await session.execute(stmt))
             except Exception:
                 # TODO: temporary workaround until we can reproduce the error
                 return []

phoenix/server/api/types/GenerativeModel.py

@@ -6,14 +6,19 @@ import strawberry
 from openinference.semconv.trace import OpenInferenceLLMProviderValues
 from sqlalchemy import inspect
 from strawberry.relay import Node, NodeID
+from strawberry.relay.types import GlobalID
 from strawberry.types import Info
-from typing_extensions import assert_never
+from strawberry.types.unset import UNSET
+from typing_extensions import TypeAlias, assert_never
 
 from phoenix.db import models
 from phoenix.server.api.context import Context
+from phoenix.server.api.exceptions import BadRequest
+from phoenix.server.api.input_types.TimeRange import TimeRange
 from phoenix.server.api.types.CostBreakdown import CostBreakdown
 from phoenix.server.api.types.GenerativeProvider import GenerativeProviderKey
 from phoenix.server.api.types.ModelInterface import ModelInterface
+from phoenix.server.api.types.node import from_global_id
 from phoenix.server.api.types.SpanCostDetailSummaryEntry import SpanCostDetailSummaryEntry
 from phoenix.server.api.types.SpanCostSummary import SpanCostSummary
 from phoenix.server.api.types.TokenPrice import TokenKind, TokenPrice
@@ -25,6 +30,11 @@ class GenerativeModelKind(Enum):
     BUILT_IN = "BUILT_IN"
 
 
+ProjectId: TypeAlias = int
+TimeRangeKey: TypeAlias = tuple[Optional[datetime], Optional[datetime]]
+CachedCostSummaryKey: TypeAlias = tuple[Optional[ProjectId], TimeRangeKey]
+
+
 @strawberry.type
 class GenerativeModel(Node, ModelInterface):
     id_attr: NodeID[int]
@@ -37,6 +47,18 @@ class GenerativeModel(Node, ModelInterface):
     provider_key: Optional[GenerativeProviderKey]
     costs: strawberry.Private[Optional[list[models.TokenPrice]]] = None
     start_time: Optional[datetime] = None
+    cached_cost_summary: strawberry.Private[
+        Optional[dict[CachedCostSummaryKey, SpanCostSummary]]
+    ] = None
+
+    def add_cached_cost_summary(
+        self, project_id: Optional[int], time_range: TimeRange, cost_summary: SpanCostSummary
+    ) -> None:
+        if self.cached_cost_summary is None:
+            self.cached_cost_summary = {}
+        time_range_key = (time_range.start, time_range.end) if time_range else (None, None)
+        cache_key = (project_id, time_range_key)
+        self.cached_cost_summary[cache_key] = cost_summary
 
     @strawberry.field
     async def token_prices(self) -> list[TokenPrice]:
@@ -55,7 +77,28 @@ class GenerativeModel(Node, ModelInterface):
         return token_prices
 
     @strawberry.field
-    async def cost_summary(self, info: Info[Context, None]) -> SpanCostSummary:
+    async def cost_summary(
+        self,
+        info: Info[Context, None],
+        project_id: Optional[GlobalID] = UNSET,
+        time_range: Optional[TimeRange] = UNSET,
+    ) -> SpanCostSummary:
+        if self.cached_cost_summary is not None:
+            time_range_key = (time_range.start, time_range.end) if time_range else (None, None)
+            project_rowid: Optional[int] = None
+            if project_id:
+                type_name, project_rowid = from_global_id(project_id)
+                if type_name != models.Project.__name__:
+                    raise BadRequest("Invalid Project ID")
+            cache_key = (project_rowid, time_range_key)
+            if cache_key in self.cached_cost_summary:
+                return self.cached_cost_summary[cache_key]
+
+        if time_range or project_id:
+            raise BadRequest(
+                "Cost summaries for specific projects or time ranges are not yet implemented"
+            )
+
         loader = info.context.data_loaders.span_cost_summary_by_generative_model
         summary = await loader.load(self.id_attr)
         return SpanCostSummary(
@@ -98,7 +141,9 @@ class GenerativeModel(Node, ModelInterface):
         return await info.context.data_loaders.last_used_times_by_generative_model_id.load(model_id)
 
 
-def to_gql_generative_model(model: models.GenerativeModel) -> GenerativeModel:
+def to_gql_generative_model(
+    model: models.GenerativeModel,
+) -> GenerativeModel:
     costs_are_loaded = isinstance(inspect(model).attrs.token_prices.loaded_value, list)
     name_pattern = model.name_pattern.pattern
     assert isinstance(name_pattern, str)