{
  "vertex_facts": [
    {
      "id": "vtx_auth_gcp_iam",
      "model_id": null,
      "model_family": "all models",
      "category": "auth differences",
      "description": "Authentication to Claude on Vertex AI uses standard Google Cloud IAM (service account credentials, Application Default Credentials, or short-lived OAuth2 access tokens obtained via gcloud/ADC), not a portable ANTHROPIC_API_KEY -- the same shape of difference as Bedrock's AWS SigV4/IAM auth relative to the direct API, but using Google's credential system instead of AWS's.",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:30",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_auth_gcp_iam"
    },
    {
      "id": "vtx_billing_gcp_native",
      "model_id": null,
      "model_family": "all models",
      "category": "billing",
      "description": "Usage of Claude via Vertex AI is billed through the customer's Google Cloud billing account (consolidated with other GCP spend), rather than through a separate Anthropic Console billing relationship -- the same cloud-native-billing pattern Bedrock offers via AWS billing.",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:30",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_billing_gcp_native"
    },
    {
      "id": "vtx_context_window_parity",
      "model_id": null,
      "model_family": "all models",
      "category": "context window",
      "description": "Context window sizes for a given Claude model generation are the same across the direct API, Bedrock, and Vertex AI -- context length is a property of the model itself, not of the hosting cloud. Vertex-specific request/response payload limits imposed by Google Cloud's own API gateway are a separate, infrastructure-level constraint layered on top, distinct from the model's context window.",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:41",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_context_window_parity"
    },
    {
      "id": "vtx_model_garden_availability",
      "model_id": null,
      "model_family": "all models",
      "category": "model availability",
      "description": "Claude models are available on Google Cloud through Vertex AI Model Garden -- the same underlying Claude models Anthropic serves directly, offered as a first-party, fully-managed model alongside Google's own Gemini models and other third-party models in the same catalog UI and API surface.",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:30",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_model_garden_availability"
    },
    {
      "id": "vtx_publisher_model_id_shape",
      "model_id": null,
      "model_family": "all models",
      "category": "model availability",
      "description": "Vertex AI addresses Claude models via \"publisher model\" resource names of the shape publishers/anthropic/models/<model-id>, distinct from both the direct Anthropic API's short public model names and AWS Bedrock's anthropic.<model-id>-v1:0 ARN-style IDs -- a third, GCP-specific ID convention.",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:30",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_publisher_model_id_shape"
    },
    {
      "id": "vtx_same_underlying_models",
      "model_id": null,
      "model_family": "all models",
      "category": "model availability",
      "description": "Whether accessed via the direct Anthropic API, AWS Bedrock, or Google Vertex AI, the underlying Claude model weights and capabilities are the same -- these are three different distribution/hosting surfaces for identical models, not three different models. This is the core parallel this site draws to subagentbedrock.com.",
      "source_url": "https://platform.claude.com/docs/en/build-with-claude/claude-on-amazon-bedrock",
      "created_at": "2026-07-02 08:29:41",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_same_underlying_models"
    },
    {
      "id": "vtx_quota_enterprise_agreements",
      "model_id": null,
      "model_family": "all models",
      "category": "quotas",
      "description": "Vertex AI Model Garden access to Claude is typically subject to Google Cloud project-level quota allocation and, for larger committed usage, an enterprise agreement negotiated jointly by the customer, Google Cloud, and Anthropic -- similar in spirit to Bedrock's Provisioned Throughput for guaranteed capacity, though the mechanics of provisioning differ (GCP quota requests vs. AWS Model Units).",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:41",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_quota_enterprise_agreements"
    },
    {
      "id": "vtx_region_availability",
      "model_id": null,
      "model_family": "all models",
      "category": "regions",
      "description": "Claude models on Vertex AI are available in a defined, model-specific subset of Google Cloud regions (varying by model generation), rather than being globally available from every GCP region by default -- customers should check the current per-model regional availability table in Google's own docs before deploying, since it changes as new models launch.",
      "source_url": "https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-claude",
      "created_at": "2026-07-02 08:29:41",
      "cite_as": "https://subagentvertex.com/api/vertex-facts/vtx_region_availability"
    }
  ]
}