Image Courtesy: Google

Google Cloud has announced a major expansion to its Vertex AI platform, giving developers and enterprises unprecedented flexibility to deploy leading proprietary AI models directly within their own Virtual Private Cloud (VPC).

This update marks a new era for Vertex AI—empowering organizations to combine the freedom of model choice with the control of private infrastructure.

A New Level of Flexibility and Control

Building advanced AI applications often means mixing specialized models for different use cases. Vertex AI now supports secure, self-deployable proprietary models from top AI partners, enabling teams to run models inside their own VPCs while meeting Google Cloud’s highest security and compliance standards.

These deployments fully respect VPC Service Controls (VPC-SC), ensuring that sensitive business data never leaves your environment. You can manage performance and cost by choosing machine types, scaling policies, and preferred regions—all with the flexibility to comply with data residency requirements and optimize latency.

Models Now Available in Vertex AI Model Garden

Developers can now discover and deploy a growing catalog of partner-built proprietary models via the Vertex AI Model Garden, a curated hub that already includes over 200 foundation models—ranging from Google’s Gemini family to open-source and third-party models.

Image Courtesy: Google

The new self-deployable models available today include:

• AI21 Labs – Jamba Large 1.6: A high-performance language model built for enterprise workloads.

• CAMB.AI – MARS7: A next-generation multilingual text-to-speech (TTS) model with voice cloning and emotional tuning.

• CSM – Cube: Transforms 2D images into detailed 3D models with precision and speed.

• Mistral AI – Codestral (25.01): Purpose-built for code generation and developer assistance.

• Qodo – Embed-1: A suite of code embedding models that enhance RAG systems and semantic search accuracy.

• Virtue AI – VirtueGuard: A guardrail model for real-time content moderation, compliance, and policy enforcement.

Coming soon:

• Contextual AI – Reranker, designed to improve Retrieval-Augmented Generation (RAG) performance.

• WRITER – Palmyra X4, an enterprise-grade LLM with 128K context, reasoning, code generation, and multimodal capabilities.

From Discovery to Deployment in Minutes

The process is streamlined for developers and enterprise teams:

1. Visit Vertex AI Model Garden and navigate to “Self-deploy partner models.”

2. Select a model and purchase a commercial license through the Google Cloud Marketplace.

3. Deploy the model directly within your VPC using the Model Garden’s one-click deployment workflow.

With pay-as-you-go pricing, teams pay only for what they use, and can apply existing committed-use discounts (CUDs) or reservations to optimize costs.

The Future of Open and Secure Enterprise AI

This launch reinforces Google Cloud’s mission to provide the most open, flexible, and secure AI platform for enterprises. By combining third-party innovation with the reliability of Google Cloud infrastructure, organizations can accelerate AI development while maintaining full control over their data and operations.

Explore the new proprietary models today in the Vertex AI Model Garden and start building your next-generation AI applications with confidence.