Starting April 29, 2025, Gemini 1.5 Pro and Gemini 1.5 Flash models are not available in projects that have no prior usage of these models, including new projects. For details, see Model versions and lifecycle.

Vertex AI partner models for MaaS

This document explains how to use partner models on Vertex AI as a managed service and covers the following topics:

Partner models: Learn about the partner models available as managed APIs.
Vertex AI partner model pricing with capacity assurance: Understand pricing options, including provisioned throughput for reserved capacity.
Regional and global endpoints: Choose between regional and global endpoints for your model requests.
Grant user access to partner models: Set up the necessary IAM permissions and organization policies to access partner models.

Vertex AI supports a curated list of models from Google partners, offered as a model as a service (MaaS). When you use a partner model, you send requests to a Vertex AI API endpoint, and the model runs without you needing to provision or manage any infrastructure.

You can explore and deploy partner models in Model Garden. For more information, see Explore AI models in Model Garden. This document focuses on partner models available as a MaaS on Vertex AI. For details about a specific model, see its model card in Model Garden.

Anthropic's Claude and Mistral models are examples of third-party managed models that are available to use on Vertex AI.

Partner models

The following partner models are offered as managed APIs on Vertex AI Model Garden (MaaS):

Model name	Modality	Description	Quickstart
Claude Opus 4.1	Language, Vision	An industry leader for coding. It delivers sustained performance on long-running tasks that require focused effort and thousands of steps, significantly expanding what AI agents can solve. Ideal for powering frontier agent products and features.	Model card
Claude Opus 4	Language, Vision	Claude Opus 4 delivers sustained performance on long-running tasks that require focused effort and thousands of steps, significantly expanding what AI agents can solve.	Model card
Claude Sonnet 4	Language, Vision	Anthropic's mid-size model with superior intelligence for high-volume uses, such as coding, in-depth research, and agents.	Model card
Anthropic's Claude 3.7 Sonnet	Language, Vision	Industry-leading model for coding and powering AI agents—and the first Claude model to offer extended thinking.	Model card
Anthropic's Claude 3.5 Sonnet v2	Language, Vision	The upgraded Claude 3.5 Sonnet is a state-of-the-art model for real-world software engineering tasks and agentic capabilities. Claude 3.5 Sonnet delivers these advancements at the same price and speed as its predecessor.	Model card
Anthropic's Claude 3.5 Haiku	Language, Vision	Claude 3.5 Haiku, the next generation of Anthropic's fastest and most cost-effective model, is optimal for use cases where speed and affordability matter.	Model card
Anthropic's Claude 3 Opus	Language	A powerful AI model, with top-level performance on highly complex tasks. It can navigate open-ended prompts and sight-unseen scenarios with remarkable fluency and human-like understanding.	Model card
Anthropic's Claude 3 Haiku	Language	Anthropic's fastest vision and text model for near-instant responses to basic queries, meant for seamless AI experiences mimicking human interactions.	Model card
Anthropic's Claude 3.5 Sonnet	Language	Claude 3.5 Sonnet outperforms Anthropic's Claude 3 Opus on a wide range of Anthropic's evaluations with the speed and cost of Anthropic's mid-tier model, Claude 3 Sonnet.	Model card
DeepSeek R1 (0528) (Preview)	Language	DeepSeek's latest version of the DeepSeek R1 model.	Model card
Jamba 1.5 Large (Preview)	Language	AI21 Labs's Jamba 1.5 Large is designed for superior quality responses, high throughput, and competitive pricing compared to other models in its size class.	Model card
Jamba 1.5 Mini (Preview)	Language	AI21 Labs's Jamba 1.5 Mini is well balanced across quality, throughput, and low cost.	Model card
Llama 4 Maverick 17B-128E (GA)	Language, Vision	The largest and most capable Llama 4 model that has coding, reasoning, and image capabilities. Llama 4 Maverick 17B-128E is a multimodal model that uses the Mixture-of-Experts (MoE) architecture and early fusion.	Model card
Llama 4 Scout 17B-16E (GA)	Language, Vision	Llama 4 Scout 17B-16E delivers state-of-the-art results for its size class, outperforming previous Llama generations and other open and proprietary models on several benchmarks. Llama 4 Scout 17B-16E is a multimodal model that uses the Mixture-of-Experts (MoE) architecture and early fusion.	Model card
Llama 3.3 (GA)	Language	Llama 3.3 is a text-only 70B instruction-tuned model that provides enhanced performance relative to Llama 3.1 70B and to Llama 3.2 90B when used for text-only applications. Moreover, for some applications, Llama 3.3 70B approaches the performance of Llama 3.1 405B.	Model card
Llama 3.2 (Preview)	Language, Vision	A medium-sized 90B multimodal model that can support image reasoning, such as chart and graph analysis as well as image captioning.	Model card
Llama 3.1 (GA and Preview)	Language	A collection of multilingual LLMs optimized for multilingual dialogue use cases and outperform many of the available open source and closed chat models on common industry benchmarks. Llama 3.1 405B is generally available (GA) and priced as per dollar-per-1M-tokens. See pricing. Llama 3.1 8B and Llama 3.1 70B are in Preview at no cost.	Model card
Mistral OCR (25.05)	Language, Vision	Mistral OCR (25.05) is an Optical Character Recognition API for document understanding. The model comprehends each element of documents such as media, text, tables, and equations.	Model card
Mistral Small 3.1 (25.03)	Language	Mistral Small 3.1 (25.03) is the latest version of Mistral's Small model, featuring multimodal capabilities and extended context length.	Model card
Mistral Large (24.11)	Language	Mistral Large (24.11) is the next version of the Mistral Large (24.07) model now with improved reasoning and function calling capabilities.	Model card
Codestral (25.01)	Code	A cutting-edge model that's designed for code generation, including fill-in-the-middle and code completion.	Model card

Vertex AI partner model pricing with capacity assurance

For some partner models, you can use provisioned throughput to reserve processing capacity for a fixed fee. You decide on the throughput capacity and in which regions to reserve that capacity.

Because provisioned throughput requests are prioritized over standard pay-as-you-go requests, this service provides increased availability. When the system is overloaded, your requests are still processed as long as the throughput is within your reserved capacity. For more information or to subscribe to the service, Contact sales.

Regional and global endpoints

You can send requests to partner models using either regional or global endpoints.

Endpoint Type	Description	Pros	Cons
Regional	Requests are served from the specific region you designate.	Helps meet data residency requirements.	Availability is tied to a single region.
Global	Google Cloud can process and serve requests from any region supported by the model.	Improves overall availability and can help reduce errors.	Might result in higher latency in some cases. Not all models or features (like provisioned throughput) are supported.

There is no price difference between regional and global endpoints. However, the global endpoint quotas and supported model capabilities can differ from the regional endpoints. For more information, view the related third-party model page.

Specify the global endpoint

To use the global endpoint, set the region to global.

For example, the request URL for a curl command uses the following format: https://aiplatform.googleapis.com/v1/projects/PROJECT_ID/locations/global/publishers/PUBLISHER_NAME/models/MODEL_NAME

For the Vertex AI SDK, a regional endpoint is the default. To use the global endpoint, set the region to GLOBAL.

Supported models

The global endpoint is available for the following models:

Restrict global API endpoint usage

To enforce the use of regional endpoints, use the constraints/gcp.restrictEndpointUsage organization policy constraint to block requests to the global API endpoint. For more information, see Restricting endpoint usage.

Grant user access to partner models

To enable partner models and make prompt requests, a Google Cloud administrator must set the required permissions and verify that the organization policy allows the use of required APIs.

Set required permissions to use partner models

The following roles and permissions are required to use partner models:

You must have the Consumer Procurement Entitlement Manager Identity and Access Management (IAM) role. Anyone who's been granted this role can enable partner models in Model Garden.
You must have the aiplatform.endpoints.predict permission. This permission is included in the Vertex AI User IAM role. For more information, see Vertex AI User and Access control.

Console

To grant the required IAM roles to a user, go to the IAM page.

Go to IAM
In the Principal column, find the user principal for which you want to enable access to partner models, and then click Edit principal in that row.
In the Edit access pane, click Add another role.
In Select a role, select Consumer Procurement Entitlement Manager.
In the Edit access pane, click Add another role.
In Select a role, select Vertex AI User.
Click Save.

gcloud

In the Google Cloud console, activate Cloud Shell.

Activate Cloud Shell

Grant the Consumer Procurement Entitlement Manager role that's required to enable partner models in Model Garden.

gcloud projects add-iam-policy-binding  PROJECT_ID \
--member=PRINCIPAL --role=roles/consumerprocurement.entitlementManager

Grant the Vertex AI User role that includes the aiplatform.endpoints.predict permission which is required to make prompt requests:
```
gcloud projects add-iam-policy-binding  PROJECT_ID \
--member=PRINCIPAL --role=roles/aiplatform.user
```
Replace PRINCIPAL with the identifier for the principal. The identifier takes the form user|group|serviceAccount:email or domain:domain—for example, user:cloudysanfrancisco@gmail.com, group:admins@example.com, serviceAccount:test123@example.domain.com, or domain:example.domain.com.

The output is a list of policy bindings that includes the following:
```
-   members:
  -   user:PRINCIPAL
  role: roles/roles/consumerprocurement.entitlementManager
```
For more information, see Grant a single role and gcloud projects add-iam-policy-binding.

Set the organization policy for partner model access

To enable partner models, your organization policy must allow the following API: Cloud Commerce Consumer Procurement API - cloudcommerceconsumerprocurement.googleapis.com

If your organization sets an organization policy to restrict service usage, an organization administrator must verify that cloudcommerceconsumerprocurement.googleapis.com is allowed by setting the organization policy.

Also, if you have an organization policy that restricts model usage in Model Garden, the policy must allow access to partner models. For more information, see Control model access.

Partner model regulatory compliance

The certifications for Generative AI on Vertex AI continue to apply when you use partner models as a managed API using Vertex AI. If you need details about the models themselves, you can find more information in the respective Model Card, or you can contact the model publisher.

Your data is stored at rest within the selected region or multi-region for partner models on Vertex AI, but the regionalization of data processing might vary. For a detailed list of partner models' data processing commitments, see Data residency for partner models.

When you use the Vertex AI API, including partner models, your prompts and the model responses are not shared with third-parties. Google Cloud processes Customer Data only as instructed by the Customer. For more information, see our Cloud Data Processing Addendum.