Google Cloud Vertex AI now provides access to roughly 150 open-source models through its Model Garden and Model Registry. The catalog supports enterprise teams seeking managed inference, fine-tuning, and self-deployment routes on Google infrastructure as of November 2026.
Current Catalog and Recent Additions
The registry lists between 149 and 156 open-source entries depending on the console view. Additions in recent months include multiple Qwen3.5 variants, MiniMax-M2.5, Kimi-K2.5, SAM 3, GLM-OCR, FLUX.2, several DeepSeek and Nemotron models, and Muse models from Meta. These join earlier open-weight entries sourced from Hugging Face and partner repositories.
Enterprise users can filter listings by license, modality, and deployment type. The platform supplies optimized containers such as vLLM and Hex-LLM, along with security scans before deployment. Model Registry tools handle versioning, evaluation, endpoint creation, and batch inference within the same interface.
Deployment choices include one-click managed endpoints, Model-as-a-Service APIs, notebook-based fine-tuning, GKE clusters, and Provisioned Throughput reservations. Certain models also support self-deployment inside customer VPCs.
Context for Enterprise Teams
Vertex AI Model Garden has expanded from earlier catalogs that held more than 150 total models in 2024. The service now integrates with the Gemini Enterprise Agent Platform and includes closed-source options from OpenAI and Anthropic alongside the open-source selection. This combination gives teams a single console for model discovery, comparison, and lifecycle management.
Practical use cases include fine-tuning open models on proprietary data, running batch inference jobs on existing GKE workloads, and routing production traffic through managed endpoints with usage-based pricing. Teams evaluating multiple model families can compare latency, cost, and accuracy metrics directly in the registry before committing resources.
Continued incremental updates are expected through official Google Cloud channels. Organizations monitoring release notes can track new open-weight entries and enhanced deployment features as they become available.
Frequently asked questions
How many open-source models are currently listed in Google Cloud Vertex AI?
Console listings show between 149 and 156 open-source models, with the exact number varying slightly by filter and date.
What deployment methods does Vertex AI support for open-source models?
Options include managed endpoints, Model-as-a-Service APIs, fine-tuning pipelines, GKE clusters, and self-deployment in customer VPCs.
Which recent open-source models were added to the registry?
Recent additions include Qwen3.5 variants, MiniMax-M2.5, Kimi-K2.5, SAM 3, GLM-OCR, FLUX.2, DeepSeek, Nemotron, and Muse models.
Does the Vertex AI registry include closed-source models as well?
Yes, the catalog also contains Google Gemini models, OpenAI models, and Anthropic Claude alongside the open-source selections.
By YasTech Editorial
Comments 1