diff --git a/web/app/components/evaluation/__tests__/index.spec.tsx b/web/app/components/evaluation/__tests__/index.spec.tsx
index 3d1ae351d3..a3b0d2fcac 100644
--- a/web/app/components/evaluation/__tests__/index.spec.tsx
+++ b/web/app/components/evaluation/__tests__/index.spec.tsx
@@ -6,7 +6,7 @@ import ConditionsSection from '../components/conditions-section'
 import { useEvaluationStore } from '../store'
 
 const mockUpload = vi.hoisted(() => vi.fn())
-const mockUseAvailableEvaluationMetrics = vi.hoisted(() => vi.fn())
+const mockUseDatasetEvaluationMetrics = vi.hoisted(() => vi.fn())
 const mockUseDefaultEvaluationMetrics = vi.hoisted(() => vi.fn())
 const mockUseEvaluationConfig = vi.hoisted(() => vi.fn())
 const mockUseSaveEvaluationConfigMutation = vi.hoisted(() => vi.fn())
@@ -51,7 +51,7 @@ vi.mock('@/service/base', () => ({
 
 vi.mock('@/service/use-evaluation', () => ({
   useEvaluationConfig: (...args: unknown[]) => mockUseEvaluationConfig(...args),
-  useAvailableEvaluationMetrics: (...args: unknown[]) => mockUseAvailableEvaluationMetrics(...args),
+  useDatasetEvaluationMetrics: (...args: unknown[]) => mockUseDatasetEvaluationMetrics(...args),
   useDefaultEvaluationMetrics: (...args: unknown[]) => mockUseDefaultEvaluationMetrics(...args),
   useSaveEvaluationConfigMutation: (...args: unknown[]) => mockUseSaveEvaluationConfigMutation(...args),
   useStartEvaluationRunMutation: (...args: unknown[]) => mockUseStartEvaluationRunMutation(...args),
@@ -119,7 +119,7 @@ describe('Evaluation', () => {
       data: null,
     })
 
-    mockUseAvailableEvaluationMetrics.mockReturnValue({
+    mockUseDatasetEvaluationMetrics.mockReturnValue({
       data: {
         metrics: ['answer-correctness', 'faithfulness', 'context-precision', 'context-recall', 'context-relevance'],
       },
@@ -582,6 +582,7 @@ describe('Evaluation', () => {
   it('should render the pipeline-specific layout without auto-selecting a judge model', () => {
     renderWithQueryClient(<Evaluation resourceType="datasets" resourceId="dataset-1" />)
 
+    expect(mockUseDatasetEvaluationMetrics).toHaveBeenCalledWith('dataset-1')
     expect(screen.getByTestId('evaluation-model-selector')).toHaveTextContent('empty')
     expect(screen.getByText('evaluation.history.columns.time')).toBeInTheDocument()
     expect(screen.getByText('Context Precision')).toBeInTheDocument()
diff --git a/web/app/components/evaluation/components/pipeline/pipeline-metrics-section.tsx b/web/app/components/evaluation/components/pipeline/pipeline-metrics-section.tsx
index 553083f867..568b66b5a6 100644
--- a/web/app/components/evaluation/components/pipeline/pipeline-metrics-section.tsx
+++ b/web/app/components/evaluation/components/pipeline/pipeline-metrics-section.tsx
@@ -7,7 +7,7 @@ import { useEffect, useMemo } from 'react'
 import { useTranslation } from 'react-i18next'
 import { BlockEnum } from '@/app/components/workflow/types'
 import { useDatasetDetailContextWithSelector } from '@/context/dataset-detail'
-import { useAvailableEvaluationMetrics } from '@/service/use-evaluation'
+import { useDatasetEvaluationMetrics } from '@/service/use-evaluation'
 import { usePublishedPipelineInfo } from '@/service/use-pipeline'
 import { useEvaluationResource, useEvaluationStore } from '../../store'
 import { buildMetricOption } from '../metric-selector/utils'
@@ -49,7 +49,7 @@ const PipelineMetricsSection = ({
   const addBuiltinMetric = useEvaluationStore(state => state.addBuiltinMetric)
   const removeMetric = useEvaluationStore(state => state.removeMetric)
   const updateMetricThreshold = useEvaluationStore(state => state.updateMetricThreshold)
-  const { data: availableMetricsData } = useAvailableEvaluationMetrics()
+  const { data: datasetMetricsData } = useDatasetEvaluationMetrics(resourceId)
   const { data: publishedPipeline } = usePublishedPipelineInfo(pipelineId || '')
   const resource = useEvaluationResource(resourceType, resourceId)
   const knowledgeIndexNodeInfoList = useMemo(
@@ -63,12 +63,12 @@ const PipelineMetricsSection = ({
   ), [resource.metrics])
   const availableBuiltinMetrics = useMemo(() => {
     const metricIds = new Set([
-      ...(availableMetricsData?.metrics ?? []),
+      ...(datasetMetricsData?.metrics ?? []),
       ...builtinMetricMap.keys(),
     ])
 
     return Array.from(metricIds).map(metricId => buildMetricOption(metricId))
-  }, [availableMetricsData?.metrics, builtinMetricMap])
+  }, [datasetMetricsData?.metrics, builtinMetricMap])
 
   useEffect(() => {
     if (!knowledgeIndexNodeInfoList.length)
diff --git a/web/contract/console/evaluation.ts b/web/contract/console/evaluation.ts
index 47580da438..ac961732f8 100644
--- a/web/contract/console/evaluation.ts
+++ b/web/contract/console/evaluation.ts
@@ -284,13 +284,6 @@ export const evaluationNodeInfoContract = base
   }>())
   .output(type<EvaluationNodeInfoResponse>())
 
-export const availableEvaluationMetricsContract = base
-  .route({
-    path: '/evaluation/available-metrics',
-    method: 'GET',
-  })
-  .output(type<EvaluationMetricsListResponse>())
-
 export const availableEvaluationWorkflowsContract = base
   .route({
     path: '/workspaces/current/available-evaluation-workflows',
diff --git a/web/contract/router.ts b/web/contract/router.ts
index dc00a5cbbb..b02071ab28 100644
--- a/web/contract/router.ts
+++ b/web/contract/router.ts
@@ -3,7 +3,6 @@ import { accountAvatarContract } from './console/account'
 import { appDeleteContract, appWorkflowTypeConvertContract, workflowOnlineUsersContract } from './console/apps'
 import { bindPartnerStackContract, invoicesContract } from './console/billing'
 import {
-  availableEvaluationMetricsContract,
   availableEvaluationWorkflowsContract,
   cancelDatasetEvaluationRunContract,
   cancelEvaluationRunContract,
@@ -148,7 +147,6 @@ export const consoleRouterContract = {
     metrics: evaluationMetricsContract,
     defaultMetrics: evaluationDefaultMetricsContract,
     nodeInfo: evaluationNodeInfoContract,
-    availableMetrics: availableEvaluationMetricsContract,
     availableWorkflows: availableEvaluationWorkflowsContract,
     associatedTargets: evaluationWorkflowAssociatedTargetsContract,
     file: evaluationFileContract,
diff --git a/web/service/use-evaluation.ts b/web/service/use-evaluation.ts
index cb3f46bf5b..84750417c3 100644
--- a/web/service/use-evaluation.ts
+++ b/web/service/use-evaluation.ts
@@ -54,9 +54,17 @@ export const useEvaluationConfig = (
   return useQuery<EvaluationConfig>(getEvaluationConfigQueryOptions(resourceType, resourceId))
 }
 
-export const useAvailableEvaluationMetrics = (enabled = true) => {
-  return useQuery(consoleQuery.evaluation.availableMetrics.queryOptions({
-    enabled,
+export const useDatasetEvaluationMetrics = (datasetId: string, enabled = true) => {
+  return useQuery(consoleQuery.datasetEvaluation.metrics.queryOptions({
+    input: datasetId
+      ? {
+          params: {
+            datasetId,
+          },
+        }
+      : skipToken,
+    enabled: !!datasetId && enabled,
+    refetchOnWindowFocus: false,
   }))
 }