Claude 3.7 Sonnet on Vertex AI Is Retired | Migration Targets and Current Models

Can You Still Use Claude 3.7 Sonnet on Vertex AI? — The Bottom Line

The bottom line: as of June 2026, Claude 3.7 Sonnet is explicitly listed as "Retired" in the Vertex AI model catalog and cannot be called for new requests. The Vertex AI model ID is claude-3-7-sonnet@20250219, but requests to a retired model will fail.

The most important point to understand here is that Vertex AI is not a platform operated by Anthropic itself — it is a partner platform operated by Google Cloud. As explained below, the retirement schedule is managed separately from the Anthropic direct API, so applying information about when it was retired "on the direct API" directly to Vertex will lead you astray.

If you are building a new system, there is no reason to choose 3.7 Sonnet. The right approach is to adopt its direct successor, Claude Sonnet 4.6 (claude-sonnet-4-6), from the start.

Source: Claude on Vertex AI — Claude API Docs

What Is Claude 3.7 Sonnet — Why It Attracted Attention on Vertex

Claude 3.7 Sonnet is the "world's first hybrid reasoning model," announced by Anthropic in February 2025. What made it groundbreaking was its ability to switch, depending on the use case, between answering questions instantly and reasoning through each step (extended thinking) while showing that process to the user. Its significantly improved coding performance also earned it strong support as a model built with real-world use in mind.

Google Cloud announced this model on February 25, 2025, and made it generally available (GA) on Vertex AI on March 18, 2025. The key features enhanced on the Vertex AI side at the time were as follows.

1. Token counting (GA): Check the token count of a message before sending a request, useful for estimating prompts and costs.
2. Citations (GA): Returns specific sentences and paragraphs that the response is based on as references, improving the verifiability of output.
3. Batch prediction (Preview): Process large numbers of requests asynchronously at 50% lower cost than standard pay-as-you-go pricing.
4. Prompt caching (Preview): Cache frequently occurring inputs to reduce costs.

Access was enabled from the Claude 3.7 Sonnet model card in Vertex AI Model Garden, and procurement through the Google Cloud Marketplace could be applied toward Google Cloud spending commitments.

Source: Anthropic's Claude 3.7 Sonnet is available on Vertex AI — Google Cloud Blog

The Current Status of 3.7 in the Vertex AI Model Catalog

Looking at the list of API model IDs for Claude models currently available on Vertex AI makes the situation with the 3.7 generation clear. One key thing to note is that the Vertex AI model ID uses an @ separator (claude-3-7-sonnet@20250219) rather than the hyphen separator used in the direct API (claude-3-7-sonnet-20250219).

In the catalog, Claude 3.7 Sonnet is clearly annotated as "Retired." This means it is no longer available — a status distinct from Deprecated (not recommended but still functional). For example, in the same catalog, Claude Sonnet 4 (claude-sonnet-4@20250514) and Claude Haiku 3.5 (claude-3-5-haiku@20241022) remain with a "Deprecated" annotation, but 3.7 Sonnet has moved beyond that to fully retired.

If you want to check the latest status in your own environment, search for "Claude" in the Vertex AI Model Garden, or check the retirement dates for each model in the Google Cloud documentation for Claude models.

Source: Anthropic's Claude models on Vertex AI — Google Cloud

Why Do Retirement Dates Differ Between Vertex AI and the Anthropic Direct API?

Anthropic's official deprecation documentation includes a clear note. It states that "the dates on this page apply to platforms operated by Anthropic (the Claude API, Claude Platform on AWS, and Microsoft Foundry). Partner-operated platforms such as Amazon Bedrock and Vertex AI set their own retirement schedules, so the lifecycle status and dates for models may differ."

The retirement dates for Claude 3.7 Sonnet (claude-3-7-sonnet-20250219) on the Anthropic direct API are as follows.

1. Deprecation (marking as not recommended) was announced on October 28, 2025.
2. It was retired on February 19, 2026, after which requests began to fail.
3. The recommended replacement is Claude Sonnet 4.6 (claude-sonnet-4-6).

These are the dates for the direct API only. Since the exact retirement date on Vertex AI is managed by Google Cloud, it is important not to assume that "it retired in February on the direct API, so it must be the same date on Vertex" — always verify in the model catalog for the platform you are using.

Source: Model deprecations — Claude API Docs

Migration Target — Switching to Claude Sonnet 4.6

To migrate from Claude 3.7 Sonnet, switch to the current model on Vertex AI. The direct replacement officially recommended by Anthropic is Claude Sonnet 4.6 (Vertex model ID: claude-sonnet-4-6). As the mid-tier model successor, it offers improved reasoning and coding performance, and extended thinking — popular in 3.7 — remains available.

The benefits of migrating go beyond performance. On Vertex AI, Claude Opus 4.8 / 4.7 / 4.6 and Sonnet 4.6 support a 1 million token context window (other models such as Sonnet 4.5 support 200K tokens). This is a major expansion from the 200K tokens of the 3.7 generation, enabling you to handle large codebases and long documents in a single pass.

In most cases, the migration simply involves replacing the model ID, but be aware that the handling of some parameters, such as temperature, may differ between models. Test before applying to production. If you want to understand the positioning of Claude 3.7 Sonnet itself or how it differs from OpenAI models, see Claude 3.7 Sonnet Is Made by Anthropic | Differences from OpenAI o1 as well.

Source: Model deprecations — Claude API Docs

Steps for Calling Current Claude Models on Vertex AI

Here are the quickest steps for calling the current model — migrated from 3.7 Sonnet — on Vertex AI.

1. Enable the Vertex AI API in your GCP project and enable the target Claude model in Model Garden.
2. Run gcloud auth application-default login locally to set up Google Cloud credentials.
3. Install the SDK (Python example: pip install -U google-cloud-aiplatform "anthropic[vertex]"; TypeScript: npm install @anthropic-ai/vertex-sdk).
4. Initialize an AnthropicVertex client with your project_id and region (recommended: global), specify the current model ID, and send a message.
5. If calling via REST directly, include model in the endpoint URL rather than the request body, and always specify anthropic_version: "vertex-2023-10-16" in the body.

A minimal Python sample is shown below.

from anthropic import AnthropicVertex

client = AnthropicVertex(project_id="MY_PROJECT_ID", region="global")

message = client.messages.create(
    model="claude-sonnet-4-6",  # 旧: claude-3-7-sonnet@20250219
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hey Claude!"}],
)
print(message)

The endpoint format is https://{LOCATION}-aiplatform.googleapis.com/v1/projects/{PROJECT_ID}/locations/{LOCATION}/publishers/anthropic/models/{MODEL_ID}:streamRawPredict. Vertex AI has three types of endpoints — global, multi-region, and regional — and regional and multi-region endpoints are 10% more expensive than global. If you have no data residency requirements, global is recommended. For more detailed steps on enabling and authenticating, see How to Use the Claude API with Google | Vertex AI Steps and Pricing as well.

Source: Claude on Vertex AI — Claude API Docs

Frequently Asked Questions (FAQ)

Q. What happens if I specify claude-3-7-sonnet@20250219 on Vertex AI? A. Since Claude 3.7 Sonnet is "Retired" on Vertex AI, new calls will fail. Replace it with a current model such as claude-sonnet-4-6.

Q. I was using 3.7's extended thinking. Can I still use it with the successor? A. Yes. Extended thinking is supported in the successor, Claude Sonnet 4.6, with better performance than in the 3.7 era.

Q. How much will my code need to change for the migration? A. In most cases, just replacing the model ID will work. The key things to know for Vertex AI are that the ID for the latest generation does not use @ separators (e.g., claude-sonnet-4-6), and that the anthropic_version: vertex-2023-10-16 specification remains unchanged. Confirm any parameter specification differences, such as temperature, through testing.

Q. What is the status of Claude 3.5 on Vertex? A. Claude 3.5 Sonnet has also been removed from the catalog, leaving an asymmetric situation where only Haiku 3.5 survives as Deprecated. For details, see Can You Use Claude 3.5 on Vertex AI | Retirement Status and Migration Targets.

Whether building a new system or extending the life of an existing one, there is almost no reason to choose Claude 3.7 Sonnet as of 2026. Migrating sooner rather than later to Claude Sonnet 4.6 — the current model on Vertex AI — is the most rational choice across performance, context length, and support.

Sources:

• Claude on Vertex AI — Claude API Docs
• Model deprecations — Claude API Docs
• Anthropic's Claude 3.7 Sonnet is available on Vertex AI — Google Cloud Blog