Tool tree
Workload-specific MLPerf Inference v6.0 accelerator comparisons with source-level provenance.
Implementation
gradeNot graded
gradeNot graded
list_inference_slicesAnswer
List exact MLPerf v6.0 Closed comparison slices available on this page. Call this before configuring or comparing hardware so incompatible workloads, scenarios, accuracy targets, metrics, and units are never mixed.
View tool JSON
{
"name": "list_inference_slices",
"kind": "answer",
"impl": "imperative",
"description": "List exact MLPerf v6.0 Closed comparison slices available on this page. Call this before configuring or comparing hardware so incompatible workloads, scenarios, accuracy targets, metrics, and units are never mixed.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
},
"page": "/leaderboard"
}configure_inference_leaderboardAction
Change the human-visible leaderboard to one exact slice and optional vendor, submitter, metric view, grouping, or search filter. The page updates immediately so person and agent share the same view.
View tool JSON
{
"name": "configure_inference_leaderboard",
"kind": "act",
"impl": "imperative",
"description": "Change the human-visible leaderboard to one exact slice and optional vendor, submitter, metric view, grouping, or search filter. The page updates immediately so person and agent share the same view.",
"inputSchema": {
"type": "object",
"properties": {
"sliceId": {
"type": "string",
"description": "Exact slice ID returned by list_inference_slices."
},
"vendor": {
"type": "string",
"enum": [
"All",
"AMD",
"Intel",
"NVIDIA"
]
},
"submitter": {
"type": "string"
},
"metricView": {
"type": "string",
"enum": [
"official",
"derived"
]
},
"grouping": {
"type": "string",
"enum": [
"best-per-accelerator",
"all-systems"
]
},
"search": {
"type": "string",
"maxLength": 120
}
},
"required": [
"sliceId"
],
"additionalProperties": false
},
"page": "/leaderboard"
}compare_inference_acceleratorsAction
Compare 2 to 8 accelerator slugs inside one exact MLPerf slice. Returns the best verified records, missing-evidence reasons, immutable source URLs, and renders the comparison for the person on the page.
View tool JSON
{
"name": "compare_inference_accelerators",
"kind": "act",
"impl": "imperative",
"description": "Compare 2 to 8 accelerator slugs inside one exact MLPerf slice. Returns the best verified records, missing-evidence reasons, immutable source URLs, and renders the comparison for the person on the page.",
"inputSchema": {
"type": "object",
"properties": {
"sliceId": {
"type": "string",
"description": "Exact slice ID returned by list_inference_slices."
},
"acceleratorSlugs": {
"type": "array",
"items": {
"type": "string"
},
"minItems": 2,
"maxItems": 8,
"uniqueItems": true
},
"metricView": {
"type": "string",
"enum": [
"official",
"derived"
]
}
},
"required": [
"sliceId",
"acceleratorSlugs"
],
"additionalProperties": false
},
"page": "/leaderboard"
}clear_inference_comparisonAction
Remove the agent-built comparison card from the page without changing the leaderboard filters.
View tool JSON
{
"name": "clear_inference_comparison",
"kind": "act",
"impl": "imperative",
"description": "Remove the agent-built comparison card from the page without changing the leaderboard filters.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
},
"page": "/leaderboard"
}