{
  "id": "provider:gcp-vertex",
  "_kind": "Provider",
  "_file": "compute/providers/gcp-vertex.yaml",
  "_cluster": "compute",
  "attributes": {
    "displayName": "Google Cloud Vertex AI",
    "vendor": "Google Cloud",
    "versionRange": ">=v1",
    "authMethods": [
      "service-account",
      "oauth",
      "workload-identity",
      "gcp-adc"
    ],
    "authMethodNotes": "Vertex requires GCP IAM. Local dev uses Application Default\nCredentials (`gcloud auth application-default login`); production\nworkloads use Workload Identity (GKE / Cloud Run) or service-account\nJSON. Caller MUST have `aiplatform.endpoints.predict` (and\n`aiplatform.endpoints.streamRawPredict` for streaming). The publisher\nmodel (e.g. `publishers/anthropic/models/claude-opus-4-7`) MUST be\nenabled for the GCP project.\n",
    "endpoints": {
      "base": "https://{region}-aiplatform.googleapis.com/v1",
      "publishers_anthropic_messages": "https://{region}-aiplatform.googleapis.com/v1/projects/{project}/locations/{region}/publishers/anthropic/models/{model}:rawPredict",
      "publishers_google_generate_content": "https://{region}-aiplatform.googleapis.com/v1/projects/{project}/locations/{region}/publishers/google/models/{model}:generateContent"
    },
    "pricing": "See https://cloud.google.com/vertex-ai/generative-ai/pricing for\nVertex-published model pricing. Vertex passes through model\npricing for Anthropic and Google models with a multi-region surcharge.\n",
    "pricingTiers": [
      {
        "name": "on-demand",
        "rateLimit": "Per-project online prediction quotas (RPM) per region per publisher model",
        "priceMultiplier": 1,
        "description": "Default per-token billing via :rawPredict / :generateContent."
      },
      {
        "name": "provisioned-throughput",
        "rateLimit": "Reserved GSU (Generative AI Scale Units) per region per model",
        "priceMultiplier": 1,
        "description": "Vertex Provisioned Throughput — committed capacity in GSU."
      },
      {
        "name": "batch",
        "rateLimit": "Async batch prediction jobs; backed by Cloud Storage / BigQuery",
        "priceMultiplier": 0.5,
        "description": "Vertex Batch Prediction — discounted async."
      }
    ],
    "rateLimitSignalingProtocol": "Quota errors return HTTP 429 with gRPC `code: 8 (RESOURCE_EXHAUSTED)`\nin the standard Google error envelope `{ \"error\": { \"code\": 429,\n\"status\": \"RESOURCE_EXHAUSTED\", \"message\": ..., \"details\": [...] } }`.\n`x-goog-request-id` always present. Use Cloud Quotas API to inspect\ncurrent limits.\n",
    "dataResidencyOptions": [
      "us-central1",
      "us-east1",
      "us-east4",
      "us-east5",
      "us-west1",
      "us-west4",
      "europe-west1",
      "europe-west2",
      "europe-west4",
      "europe-west9",
      "asia-southeast1",
      "asia-northeast1",
      "region:us",
      "region:eu"
    ],
    "vendorFeatures": [
      "capability:supports-vertex-reasoning-engine",
      "capability:supports-batch-api",
      "capability:supports-prompt-caching"
    ],
    "slaTier": "gcp-vertex-99.9",
    "regions": [
      "us-central1",
      "us-east1",
      "us-east4",
      "us-east5",
      "us-west1",
      "us-west4",
      "europe-west1",
      "europe-west2",
      "europe-west4",
      "europe-west9",
      "asia-southeast1",
      "asia-northeast1"
    ]
  },
  "outgoingEdges": [
    {
      "from": "provider:gcp-vertex",
      "to": "layer:2-provider",
      "kind": "realizes",
      "attributes": {}
    },
    {
      "from": "provider:gcp-vertex",
      "to": "capability:supports-vertex-reasoning-engine",
      "kind": "supports",
      "attributes": {
        "versionRange": ">=2024-01-01"
      }
    },
    {
      "from": "provider:gcp-vertex",
      "to": "capability:model-discovery",
      "kind": "supports",
      "attributes": {
        "versionRange": ">=2024-01-01"
      }
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-opus-4-7@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-opus-4-6@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-opus-4-5@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-opus-3@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-sonnet-4-6@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-sonnet-4-5@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-sonnet-3-5@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:claude-haiku-4-5@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:gemini-3-1-pro@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:gemini-2-5-pro@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:gemini-2-5-flash@current",
      "kind": "serves"
    },
    {
      "from": "provider:gcp-vertex",
      "to": "model:gemini-2-0-flash@current",
      "kind": "serves"
    }
  ],
  "incomingEdges": [
    {
      "from": "model-transport:vertex-anthropic-messages",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-haiku-4-5@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-opus-3@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-opus-4-5@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-opus-4-6@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-opus-4-7@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-sonnet-3-5@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-sonnet-4-5@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:claude-sonnet-4-6@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-2-0-flash@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-2-5-flash@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-2-5-pro@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-3-1-deep-think@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-3-1-flash-lite@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-3-1-pro@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-3-flash@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "model:gemini-3-pro@current",
      "to": "provider:gcp-vertex",
      "kind": "served_by"
    },
    {
      "from": "mux:transport-mux",
      "to": "provider:gcp-vertex",
      "kind": "bridges_for",
      "attributes": {}
    }
  ]
}