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

phoenix/server/api/types/Project.py

@@ -33,6 +33,7 @@ from phoenix.server.api.types.AnnotationConfig import AnnotationConfig, to_gql_a
 from phoenix.server.api.types.AnnotationSummary import AnnotationSummary
 from phoenix.server.api.types.CostBreakdown import CostBreakdown
 from phoenix.server.api.types.DocumentEvaluationSummary import DocumentEvaluationSummary
+from phoenix.server.api.types.GenerativeModel import GenerativeModel, to_gql_generative_model
 from phoenix.server.api.types.pagination import (
     ConnectionArgs,
     Cursor,
@@ -1326,6 +1327,130 @@ class Project(Node):
             names=sorted(list(unique_names)),
         )
 
+    @strawberry.field
+    async def top_models_by_cost(
+        self,
+        info: Info[Context, None],
+        time_range: TimeRange,
+    ) -> list[GenerativeModel]:
+        if time_range.start is None:
+            raise BadRequest("Start time is required")
+
+        async with info.context.db() as session:
+            stmt = (
+                select(
+                    models.GenerativeModel,
+                    func.sum(models.SpanCost.total_tokens).label("total_tokens"),
+                    func.sum(models.SpanCost.prompt_tokens).label("prompt_tokens"),
+                    func.sum(models.SpanCost.completion_tokens).label("completion_tokens"),
+                    func.sum(models.SpanCost.total_cost).label("total_cost"),
+                    func.sum(models.SpanCost.prompt_cost).label("prompt_cost"),
+                    func.sum(models.SpanCost.completion_cost).label("completion_cost"),
+                )
+                .join(
+                    models.SpanCost,
+                    models.SpanCost.model_id == models.GenerativeModel.id,
+                )
+                .join(
+                    models.Trace,
+                    models.SpanCost.trace_rowid == models.Trace.id,
+                )
+                .where(models.Trace.project_rowid == self.project_rowid)
+                .where(models.SpanCost.model_id.isnot(None))
+                .where(models.SpanCost.span_start_time >= time_range.start)
+                .group_by(models.SpanCost.model_id)
+                .order_by(func.sum(models.SpanCost.total_cost).desc())
+            )
+            if time_range.end:
+                stmt = stmt.where(models.SpanCost.span_start_time < time_range.end)
+            results: list[GenerativeModel] = []
+            async for (
+                model,
+                total_tokens,
+                prompt_tokens,
+                completion_tokens,
+                total_cost,
+                prompt_cost,
+                completion_cost,
+            ) in await session.stream(stmt):
+                cost_summary = SpanCostSummary(
+                    prompt=CostBreakdown(tokens=prompt_tokens, cost=prompt_cost),
+                    completion=CostBreakdown(tokens=completion_tokens, cost=completion_cost),
+                    total=CostBreakdown(tokens=total_tokens, cost=total_cost),
+                )
+                cache_time_range = TimeRange(
+                    start=time_range.start,
+                    end=time_range.end,
+                )
+                gql_model = to_gql_generative_model(model)
+                gql_model.add_cached_cost_summary(
+                    self.project_rowid, cache_time_range, cost_summary
+                )
+                results.append(gql_model)
+            return results
+
+    @strawberry.field
+    async def top_models_by_token_count(
+        self,
+        info: Info[Context, None],
+        time_range: TimeRange,
+    ) -> list[GenerativeModel]:
+        if time_range.start is None:
+            raise BadRequest("Start time is required")
+
+        async with info.context.db() as session:
+            stmt = (
+                select(
+                    models.GenerativeModel,
+                    func.sum(models.SpanCost.total_tokens).label("total_tokens"),
+                    func.sum(models.SpanCost.prompt_tokens).label("prompt_tokens"),
+                    func.sum(models.SpanCost.completion_tokens).label("completion_tokens"),
+                    func.sum(models.SpanCost.total_cost).label("total_cost"),
+                    func.sum(models.SpanCost.prompt_cost).label("prompt_cost"),
+                    func.sum(models.SpanCost.completion_cost).label("completion_cost"),
+                )
+                .join(
+                    models.SpanCost,
+                    models.SpanCost.model_id == models.GenerativeModel.id,
+                )
+                .join(
+                    models.Trace,
+                    models.SpanCost.trace_rowid == models.Trace.id,
+                )
+                .where(models.Trace.project_rowid == self.project_rowid)
+                .where(models.SpanCost.model_id.isnot(None))
+                .where(models.SpanCost.span_start_time >= time_range.start)
+                .group_by(models.SpanCost.model_id)
+                .order_by(func.sum(models.SpanCost.total_tokens).desc())
+            )
+            if time_range.end:
+                stmt = stmt.where(models.SpanCost.span_start_time < time_range.end)
+            results: list[GenerativeModel] = []
+            async for (
+                model,
+                total_tokens,
+                prompt_tokens,
+                completion_tokens,
+                total_cost,
+                prompt_cost,
+                completion_cost,
+            ) in await session.stream(stmt):
+                cost_summary = SpanCostSummary(
+                    prompt=CostBreakdown(tokens=prompt_tokens, cost=prompt_cost),
+                    completion=CostBreakdown(tokens=completion_tokens, cost=completion_cost),
+                    total=CostBreakdown(tokens=total_tokens, cost=total_cost),
+                )
+                cache_time_range = TimeRange(
+                    start=time_range.start,
+                    end=time_range.end,
+                )
+                gql_model = to_gql_generative_model(model)
+                gql_model.add_cached_cost_summary(
+                    self.project_rowid, cache_time_range, cost_summary
+                )
+                results.append(gql_model)
+            return results
+
 
 @strawberry.type
 class SpanCountTimeSeriesDataPoint:

phoenix/server/static/.vite/manifest.json

@@ -1,87 +1,87 @@
 {
-  "_components-XAeml0-1.js": {
-    "file": "assets/components-XAeml0-1.js",
+  "_components-B7lK-RgC.js": {
+    "file": "assets/components-B7lK-RgC.js",
     "name": "components",
     "imports": [
-      "_vendor-DhvamIr8.js",
-      "_pages-CPfaxiKa.js",
-      "_vendor-arizeai-4fVwwnrI.js",
-      "_vendor-codemirror-DRfFHb57.js",
+      "_vendor-uWG2jYEi.js",
+      "_pages-CnVg_GUi.js",
+      "_vendor-arizeai-CXCKGfvH.js",
+      "_vendor-codemirror-DWr46-WB.js",
       "_vendor-three-C5WAXd5r.js"
     ]
   },
-  "_pages-CPfaxiKa.js": {
-    "file": "assets/pages-CPfaxiKa.js",
+  "_pages-CnVg_GUi.js": {
+    "file": "assets/pages-CnVg_GUi.js",
     "name": "pages",
     "imports": [
-      "_vendor-DhvamIr8.js",
-      "_vendor-arizeai-4fVwwnrI.js",
-      "_components-XAeml0-1.js",
-      "_vendor-codemirror-DRfFHb57.js",
-      "_vendor-recharts-w6bSawXG.js"
+      "_vendor-uWG2jYEi.js",
+      "_vendor-arizeai-CXCKGfvH.js",
+      "_components-B7lK-RgC.js",
+      "_vendor-codemirror-DWr46-WB.js",
+      "_vendor-recharts-0Yf6lanX.js"
     ]
   },
   "_vendor-CqDb5u4o.css": {
     "file": "assets/vendor-CqDb5u4o.css",
     "src": "_vendor-CqDb5u4o.css"
   },
-  "_vendor-DhvamIr8.js": {
-    "file": "assets/vendor-DhvamIr8.js",
-    "name": "vendor",
-    "imports": [
-      "_vendor-three-C5WAXd5r.js"
-    ],
-    "css": [
-      "assets/vendor-CqDb5u4o.css"
-    ]
-  },
-  "_vendor-arizeai-4fVwwnrI.js": {
-    "file": "assets/vendor-arizeai-4fVwwnrI.js",
+  "_vendor-arizeai-CXCKGfvH.js": {
+    "file": "assets/vendor-arizeai-CXCKGfvH.js",
     "name": "vendor-arizeai",
     "imports": [
-      "_vendor-DhvamIr8.js"
+      "_vendor-uWG2jYEi.js"
     ]
   },
-  "_vendor-codemirror-DRfFHb57.js": {
-    "file": "assets/vendor-codemirror-DRfFHb57.js",
+  "_vendor-codemirror-DWr46-WB.js": {
+    "file": "assets/vendor-codemirror-DWr46-WB.js",
     "name": "vendor-codemirror",
     "imports": [
-      "_vendor-DhvamIr8.js",
-      "_vendor-shiki-CplrhwOk.js"
+      "_vendor-uWG2jYEi.js",
+      "_vendor-shiki-Caei6iKO.js"
     ]
   },
-  "_vendor-recharts-w6bSawXG.js": {
-    "file": "assets/vendor-recharts-w6bSawXG.js",
+  "_vendor-recharts-0Yf6lanX.js": {
+    "file": "assets/vendor-recharts-0Yf6lanX.js",
     "name": "vendor-recharts",
     "imports": [
-      "_vendor-DhvamIr8.js"
+      "_vendor-uWG2jYEi.js"
     ]
   },
-  "_vendor-shiki-CplrhwOk.js": {
-    "file": "assets/vendor-shiki-CplrhwOk.js",
+  "_vendor-shiki-Caei6iKO.js": {
+    "file": "assets/vendor-shiki-Caei6iKO.js",
     "name": "vendor-shiki",
     "imports": [
-      "_vendor-DhvamIr8.js"
+      "_vendor-uWG2jYEi.js"
     ]
   },
   "_vendor-three-C5WAXd5r.js": {
     "file": "assets/vendor-three-C5WAXd5r.js",
     "name": "vendor-three"
   },
+  "_vendor-uWG2jYEi.js": {
+    "file": "assets/vendor-uWG2jYEi.js",
+    "name": "vendor",
+    "imports": [
+      "_vendor-three-C5WAXd5r.js"
+    ],
+    "css": [
+      "assets/vendor-CqDb5u4o.css"
+    ]
+  },
   "index.tsx": {
-    "file": "assets/index-D7EtHUpz.js",
+    "file": "assets/index-CpePoyTU.js",
     "name": "index",
     "src": "index.tsx",
     "isEntry": true,
     "imports": [
-      "_vendor-DhvamIr8.js",
-      "_vendor-arizeai-4fVwwnrI.js",
-      "_pages-CPfaxiKa.js",
-      "_components-XAeml0-1.js",
+      "_vendor-uWG2jYEi.js",
+      "_vendor-arizeai-CXCKGfvH.js",
+      "_pages-CnVg_GUi.js",
+      "_components-B7lK-RgC.js",
       "_vendor-three-C5WAXd5r.js",
-      "_vendor-codemirror-DRfFHb57.js",
-      "_vendor-shiki-CplrhwOk.js",
-      "_vendor-recharts-w6bSawXG.js"
+      "_vendor-codemirror-DWr46-WB.js",
+      "_vendor-shiki-Caei6iKO.js",
+      "_vendor-recharts-0Yf6lanX.js"
     ]
   }
 }