How to up vote this for it to get noticed? I have similar requirement.
Hey @Surendra_Galwa, as it has been about half a year and the post went left unanswered, I assume you’d have figured something by now. How did you proceeded? Web scraping? or did you found some other resource? Please share.
Hey @ilnardo92, can you please confirm if it is intentionally not developed? or please tag in someone from the corresponding team here to guide the way or suggest an alternative method.
The requirement is:
When I visit Claude Opus 4.7 (or some other model) model card in model garden, I get a lot of information laid out about the model. Yet, when I try to fetch the same information programmatically, no way returns satisfactory amount of data.
from google.cloud.aiplatform_v1 import ModelGardenServiceClient
from google.protobuf.json_format import MessageToJson
client = ModelGardenServiceClient(
credentials=credentials, # instance of google.oauth2.service_account.Credentials
client_options={"api_endpoint": "us-central1-aiplatform.googleapis.com"}
)
publisher_model = client.get_publisher_model(
name="publishers/anthropic/models/claude-opus-4-7"
)
print(MessageToJson(publisher_model._pb, indent=4))
Above code returned this minimal data.
{
"name": "publishers/anthropic/models/claude-opus-4-7",
"versionId": "default",
"openSourceCategory": "PROPRIETARY",
"supportedActions": {
"openNotebook": {
"references": {
"us-central1": {
"uri": "https://colab.research.google.com/github/GoogleCloudPlatform/vertex-ai-samples/blob/main/notebooks/official/generative_ai/anthropic_claude_intro.ipynb"
}
},
"title": "Open Notebook"
},
"openGenerationAiStudio": {
"references": {
"us-central1": {
"uri": "https://console.cloud.google.com/vertex-ai/generative/multimodal/create/text?model=claude-opus-4-7"
}
}
},
"requestAccess": {
"references": {
"us-central1": {
"uri": "https://console.cloud.google.com/vertex-ai/model-garden/questionnaire?model=publishers/anthropic/models/claude-opus-4-7&mp=anthropic/anthropic-895.cloudpartnerservices.goog&service=anthropic-895.cloudpartnerservices.goog&enableLoggingDisclaimer=true&enableWebsearchDisclaimer=true"
}
}
},
"openNotebooks": {
"notebooks": [
{
"references": {
"us-central1": {
"uri": "https://colab.research.google.com/github/GoogleCloudPlatform/vertex-ai-samples/blob/main/notebooks/official/generative_ai/anthropic_claude_intro.ipynb"
}
},
"title": "Open Notebook"
}
]
}
},
"launchStage": "GA"
}
Whereas, on the the platform, this API get called.
curl --location 'https://cloudconsole-pa.clients6.google.com/v3/entityServices/AiplatformEntityService/schemas/AIPLATFORM_GRAPHQL:batchGraphql' \
--header 'content-type: application/json' \
--header 'origin: https://console.cloud.google.com' \
--data '{
"requestContext": {...},
"querySignature": "",
"operationName": "PublisherModel",
"variables": {
"name": "publishers/anthropic/models/claude-opus-4-7"
}
}'
That returns this data, which then is rendered on the page.
[
{
"results": [
{
"data": {
"ui": {
"publisherModel": {
"name": "publishers/anthropic/models/claude-opus-4-7",
"displayName": "Claude Opus 4.7",
"overview": "Claude Opus 4.7 is our most capable production model yet, advancing performance across coding, enterprise workflows, and long-running agentic tasks.",
"modelServingType": "MAAS_MODEL",
"categories": [
"FOUNDATION"
],
"architectures": [],
"supportedTasks": [
"GENERATION",
"EXTRACTION",
"RECOGNITION",
"DETECTION",
"TRANSLATION",
"CLASSIFICATION"
],
"inputTypes": [
"MULTIMODAL",
"LANGUAGE",
"VISION",
"DOCS",
"TABULAR"
],
"outputTypes": [
"LANGUAGE",
"TABULAR"
],
"frameworks": [],
"disclaimer": "DISCLAIMER_UNSPECIFIED",
"documentations": [
{
"title": "Overview",
"content": "\u003c!-- mdlint off(LINE_OVER_80) --\u003e\n\nClaude Opus 4.7 is our most capable production model yet, advancing performance across coding, enterprise workflows, and long-running agentic tasks.\n\n# Model Details\n\nDetails are as follows\n\n| Property | Claude Opus 4.7 |\n|:---------------------|:-----------------------------------------------------|\n| Model Name | claude-opus-4-7 |\n| Release Date | April 16, 2026 |\n| Token Limits | Inputs: 1M \u003cbr\u003e Output: 128,000 tokens |\n| Supported Data Types | Inputs: Image, PDF, Text \u003cbr\u003e Output: Text |\n"
},
{
"title": "Feature Support",
"content": "\u003c!-- mdlint off(LINE_OVER_80) --\u003e\n\nThis model includes supports the following features:\n\n- **Memory tool (beta):** Store and retrieve information across conversations for enhanced context management\n- **1M token context window:** Process and maintain coherent understanding across extended documents and\n conversations\n- **Enhanced tool orchestration:** Parallel tool execution and automatic tool call management\n- [Prompt caching](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/claude-prompt-caching)\n- [Batch prediction](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/claude-batch)\n- [Count tokens](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/claude-count-tokens)\n- [Global endpoint](https://cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-partner-models)\n- [Multi-region endpoint](https://docs.cloud.google.com/vertex-ai/generative-ai/docs/partner-models/use-partner-models#multi-region)\n"
},
{
"title": "Use Cases",
"content": "\u003c!-- mdlint off(LINE_OVER_80) --\u003e\n\n- **Coding**: Claude Opus 4.7 is built for agentic coding at scale, excelling at long-horizon projects, complex\n implementations, and polished UI design. It handles the full lifecycle from architecture to deployment, including\n design-quality UI so senior engineers can delegate complex work with confidence.\n- **Enterprise workflows**: Claude Opus 4.7 sets the standard for enterprise workflows, carrying context across\n sessions to manage complex, multi-day projects end-to-end with professional polish and industry-leading performance\n on spreadsheets, slides, and docs.\n- **Long-running agents**: Claude Opus 4.7 powers production agentic workflows, orchestrating complex multi-tool tasks\n with industry-leading reliability. It plans deliberately, uses memory to learn across sessions, and drives\n long-running work forward with minimal oversight.\n- **Financial analysis**: Claude Opus 4.7 brings frontier reasoning to financial workflows, reading dense filings and\n charts at high fidelity and carrying context across an entire deal or reporting cycle. It handles the nuance and\n precision that compliance-sensitive work demands.\n- **Cybersecurity**: Claude Opus 4.7 advances reasoning for security workflows, holding long traces and large codebases\n in context to catch subtle patterns and complex attack vectors.\n- **Computer use**: Claude Opus 4.7 is our most capable production model for computer-use, bringing high-resolution\n vision and deep reasoning to multi-step tasks that span multiple applications and require planning and judgment.\n"
},
{
"title": "Documentation",
"content": "\u003c!-- mdlint off(LINE_OVER_80) --\u003e\n\n### Quick Start (5 minutes)\n\nGet up and running with Claude Opus 4.7 in just a few steps:\n\n#### Prerequisites\n\n- [Google Cloud Project](https://console.cloud.google.com/) with billing enabled\n- [Agent Platform API enabled](https://console.cloud.google.com/marketplace/product/google/aiplatform.googleapis.com)\n- [gcloud CLI installed](https://cloud.google.com/sdk/docs/install) and authenticated\n\n#### Try it now\n\n**Quick Python test**\n\n ```bash\n pip install 'anthropic[vertex]'\n ```\n\n ```python\n from anthropic import AnthropicVertex\n\nclient = AnthropicVertex(region=\"global\", project_id=\"YOUR_PROJECT_ID\")\nmessage = client.messages.create(\n max_tokens=1024,\n messages=[{\"role\": \"user\", \"content\": \"Hello! Can you help me?\"}],\n model=\"claude-opus-4-7\"\n)\nprint(message.content[0].text)\n ```\n\nReady to build something more complex? Continue to the detailed setup below.\n\n---\n\n### Get started\n\nYou can use Claude Opus 4.7 in Agent Studio or use the API to integrate the model in your application.\n\n#### Before you begin\n\n- Enable the [Agent Platform API](https://console.cloud.google.com/marketplace/product/google/aiplatform.googleapis.com).\n- Authenticate with one of the standard mechanisms\n documented [here](https://cloud.google.com/docs/authentication/provide-credentials-adc).\n\n#### Try Claude Opus 4.7 in Agent Studio (console)\n\nTo use Claude Opus 4.7 in Agent Studio,\nvisit [Agent Studio](https://console.cloud.google.com/vertex-ai/studio/multimodal;mode=prompt?model=claude-opus-4-7)\nand select Claude Opus 4.7. You can write a prompt then click Submit to view the output generated by Claude Opus\n4.7.\n\n#### Try Claude Opus 4.7 (cURL)\n\nThe following is a sample prompt to the model. To learn more about the possible request parameters, see\nthe [Claude Messages API Reference](https://docs.claude.com/en/api/messages).\n\nNote that the API for Claude on Vertex differs from the Anthropic API documentation in the following ways:\n\n- `model` is not a valid parameter. The model is instead specified in the Google Cloud endpoint URL.\n- `anthropic_version` is a required parameter and must be set to `vertex-2023-10-16`.\n\nRequest JSON body:\n\n```json\n{\n \"anthropic_version\": \"vertex-2023-10-16\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"Send me a recipe for banana bread.\"\n }\n ],\n \"max_tokens\": 1024,\n \"stream\": true\n}\n```\n\nSet `stream` to true to\nincrementally [stream the response](https://docs.anthropic.com/claude/reference/messages-streaming) using server-sent\nevents. Streaming substantially reduces end-user perception of latency, because the response is returned incrementally\nas it's generated.\n\nReplace `your-gcp-project-id` with your actual project ID and run:\n\n```shell\ncurl -X POST \\\n -H \"Authorization: Bearer $(gcloud auth print-access-token)\" \\\n -H \"Content-Type: application/json; charset=utf-8\" \\\n -d '{\n \"anthropic_version\": \"vertex-2023-10-16\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"Send me a recipe for banana bread.\"\n }\n ],\n \"max_tokens\": 1024,\n \"stream\": true\n }' \\\n \"https://aiplatform.googleapis.com/v1/projects/your-gcp-project-id/locations/global/publishers/anthropic/models/claude-opus-4-7:streamRawPredict\"\n```\n\n**Expected Response:**\n\n```json\n{\n \"id\": \"msg_01AbCdEfGhIjKlMnOpQrStUv\",\n \"type\": \"message\",\n \"role\": \"assistant\",\n \"content\": [\n {\n \"type\": \"text\",\n \"text\": \"Here's a delicious banana bread recipe:\\n\\n**Classic Banana Bread**\\n\\nIngredients:\\n- 3-4 ripe bananas, mashed\\n- 1/3 cup melted butter\\n- 3/4 cup sugar\\n- 1 egg, beaten\\n- 1 teaspoon vanilla extract\\n- 1 teaspoon baking soda\\n- Pinch of salt\\n- 1 1/2 cups all-purpose flour\\n\\nInstructions:\\n1. Preheat oven to 350°F (175°C)\\n2. Mix melted butter with mashed bananas\\n3. Mix in sugar, egg, and vanilla\\n4. Sprinkle baking soda and salt over mixture, then add flour\\n5. Pour into buttered loaf pan\\n6. Bake for 60 minutes or until toothpick comes out clean\\n7. Cool before removing from pan\\n\\nEnjoy your homemade banana bread!\"\n }\n ],\n \"model\": \"claude-opus-4-7\",\n \"usage\": {\n \"input_tokens\": 12,\n \"output_tokens\": 184\n }\n}\n```\n\n#### Try Claude Opus 4.7 (Anthropic's Vertex SDK)\n\n[Anthropic's Vertex SDK](https://docs.claude.com/en/api/claude-on-vertex-ai) support Python and TypeScript.\n\nTo install the Anthropic Python SDK:\n\n```shell\npython -m pip install -U 'anthropic[vertex]'\n```\n\nTo use the SDK:\n\n```python\nfrom anthropic import AnthropicVertex\n\nLOCATION = \"global\"\n\nclient = AnthropicVertex(region=LOCATION, project_id=\"PROJECT_ID\")\n\nmessage = client.messages.create(\n max_tokens=1024,\n messages=[\n {\n \"role\": \"user\",\n \"content\": \"Send me a recipe for banana bread.\",\n }\n ],\n model=\"claude-opus-4-7\"\n)\nprint(message.model_dump_json(indent=2))\n```\n\nTo use the SDK to stream messages:\n\n```python\nfrom anthropic import AnthropicVertex\n\nLOCATION = \"global\"\n\nclient = AnthropicVertex(region=LOCATION, project_id=\"PROJECT_ID\")\n\nwith client.messages.stream(\n max_tokens=1024,\n messages=[\n {\n \"role\": \"user\",\n \"content\": \"Send me a recipe for banana bread.\",\n }\n ],\n model=\"claude-opus-4-7\"\n) as stream:\n for text in stream.text_stream:\n print(text)\n```\n\nTo use the SDK to process images:\n\n```python\nimport base64\nimport httpx\nfrom anthropic import AnthropicVertex\n\nLOCATION = \"global\"\n\nclient = AnthropicVertex(region=LOCATION, project_id=\"PROJECT_ID\")\n\nimage1_url = \"https://upload.wikimedia.org/wikipedia/commons/a/a7/Camponotus_flavomarginatus_ant.jpg\"\nimage1_media_type = \"image/jpeg\"\nimage1_data = base64.b64encode(httpx.get(image1_url).content).decode(\"utf-8\")\n\nmessage = client.messages.create(\n max_tokens=1024,\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image\",\n \"source\": {\n \"type\": \"base64\",\n \"media_type\": image1_media_type,\n \"data\": image1_data,\n },\n },\n {\n \"type\": \"text\",\n \"text\": \"Describe this image.\"\n }\n ],\n }\n ],\n model=\"claude-opus-4-7\"\n)\nprint(message.model_dump_json(indent=2))\n```\n"
},
{
"title": "Additional Information",
"content": "\u003c!-- mdlint off(LINE_OVER_80) --\u003e\n\n### Versions\n\n| Version Resource ID | Date released | Tentative Retirement Date | Release stage|\n|:---------------------------|:-------------------|:--------------------------|:-------------|\n| claude-opus-4-7 | April 16th, 2026 | October 16th, 2026 | General Availability (GA) |\n\n### Links\n\n#### Getting Started\n\n- [**Claude Opus 4.7 Model Card**](https://www.anthropic.com/claude-opus-4-7-system-card) - Technical details and capabilities\n- [**Anthropic's Guide to Claude Models**](https://docs.claude.com/en/docs/about-claude/models) - Model overview and comparison\n\n#### SDKs and Development\n\n- [**Anthropic's Vertex Python SDK**](https://pypi.org/project/anthropic/)\n- [**Anthropic's Vertex TypeScript SDK**](https://anthropic.com/partner/vertex-sdk-typescript)\n- [**Anthropic Code Cookbook**](https://anthropic.com/partner/anthropic-cookbook) - Example code for complex tasks\n\n#### Documentation and Guides\n\n- [**Anthropic's Documentation**](https://docs.claude.com/) - Complete API reference and guides\n- [**Extended Thinking Guide**](https://docs.anthropic.com/en/docs/build-with-claude/extended-thinking) - Advanced reasoning techniques\n- [**Prompting Resources**](https://www.anthropic.com/partner/prompt-engineering) - Tools and best practices\n- [**Prompt Library**](https://www.anthropic.com/partner/prompt-library) - Ready-to-use prompts\n- [**Interactive Tutorial**](https://www.anthropic.com/partner/peit) - Hands-on prompt engineering course\n\n#### Business and Legal\n\n- [**Customer Stories**](https://claude.com/customers) - Real-world use cases\n- [**Terms of Service**](https://www.anthropic.com/legal/commercial-terms)\n- [**Trust Portal**](https://trust.anthropic.com) - Security and compliance information\n"
}
],
"skillLevels": [
"Beginner",
"Intermediate",
"Advanced"
],
"languages": [
"English",
"French",
"Modern Standard Arabic",
"Mandarin Chinese",
"Hindi",
"Spanish",
"Russian",
"Portuguese",
"Korean",
"Japanese",
"German",
"Polish",
"Other languages"
],
"versionExternalName": "claude-opus-4-7@default",
"versionId": "default",
"license": null,
"supportedActions": {
"viewRestApi": null,
"openNotebook": {
"title": "Open Notebook",
"references": [
{
"key": "us-central1",
"value": {
"reference": "uri",
"uri": "https://colab.research.google.com/github/GoogleCloudPlatform/vertex-ai-samples/blob/main/notebooks/official/generative_ai/anthropic_claude_intro.ipynb"
}
}
]
},
"openNotebooks": {
"notebooks": [
{
"references": [
{
"key": "us-central1",
"value": {
"reference": "uri",
"uri": "https://colab.research.google.com/github/GoogleCloudPlatform/vertex-ai-samples/blob/main/notebooks/official/generative_ai/anthropic_claude_intro.ipynb"
}
}
],
"title": "Open Notebook",
"resourceTitle": null,
"resourceUseCase": null,
"resourceDescription": null,
"supportsWorkbench": null,
"colabNotebookDisabled": null
}
]
},
"createApplication": null,
"openFineTuningPipeline": null,
"createFineTuningPipeline": null,
"openPromptTuningPipeline": null,
"openGenie": null,
"requestAccess": {
"title": "",
"references": [
{
"key": "us-central1",
"value": {
"reference": "uri",
"uri": "https://console.cloud.google.com/vertex-ai/model-garden/questionnaire?model=publishers/anthropic/models/claude-opus-4-7&mp=anthropic/anthropic-895.cloudpartnerservices.goog&service=anthropic-895.cloudpartnerservices.goog&enableLoggingDisclaimer=true&enableWebsearchDisclaimer=true",
"resourceName": null
}
}
]
},
"openGenerationAiStudio": {
"title": "",
"references": [
{
"key": "us-central1",
"value": {
"reference": "uri",
"uri": "https://console.cloud.google.com/vertex-ai/generative/multimodal/create/text?model=claude-opus-4-7"
}
}
]
},
"multiTuneVertex": null,
"vmgManagedTuning": null,
"deploy": null,
"multiDeployVertex": null,
"deployMonetizedModel": {
"monetizationConfig": {
"singleTenant": null,
"marketplaceListings": [
{
"serviceId": "services/anthropic-895.cloudpartnerservices.goog",
"serviceLevel": "base",
"disableDeploymentBilling": false
}
],
"monetizableDeploymentUnits": []
},
"serviceAccount": ""
},
"deployGke": null,
"openEvaluationPipeline": null,
"batchPrediction": null
},
"projectMetadata": null,
"parent": null,
"labels": [],
"tryItOut": null,
"launchStage": "GA"
}
}
},
"path": []
}
],
"responseContext": {...}
}
]
It is not possible to call this API from code unless browser automation code is implemented. But it is a tedious task with numerous problems.
I require a way to get the list of models in garden models and their model card information programmatically.