Cortex AI Gateway expands to Azure, GCP, and new AWS regions in Snowflake
Cortex AI Gateway is now available in Azure, GCP, and AWS New Zealand commercial regions, centralizing AI model governance, spend tracking, and observability.
O que mudou
Cortex AI Gateway has expanded its availability to commercial regions across Microsoft Azure, Google Cloud Platform (GCP), and AWS New Zealand. It provides a centralized, governed access point to manage how AI clients and applications consume models within your account.
Use cases
Imagine your team operates multiple AI applications and coding agents that interact with language models, and you need a single location to attribute spend, control model access, and audit usage. Cortex AI Gateway gives platform teams a unified endpoint to enforce access rules, apply budgets, and log telemetry.
How to use
Snowflake automatically provisions a gateway object named SNOWFLAKE for every account. You do not need to create it manually; you simply manage its privileges and YAML specification.
To list gateways and retrieve the endpoint URL:
SHOW AI GATEWAYS;
What happens in this code:
- Lists all AI Gateways present in the account.
- Returns the precise endpoint URL for client applications to connect to.
To grant usage or monitoring privileges to roles:
GRANT USAGE ON AI GATEWAY SNOWFLAKE TO ROLE ai_engineering;
GRANT MONITOR ON AI GATEWAY SNOWFLAKE TO ROLE finops_admin;
What happens in this code:
- Allows the
ai_engineeringrole to send inference requests through the gateway. - Grants the
finops_adminrole access to query gateway usage metrics and history.
To inspect and edit gateway configuration:
DESCRIBE AI GATEWAY SNOWFLAKE;
ALTER AI GATEWAY SNOWFLAKE FROM SPECIFICATION $$
schema_version: 1
models:
- name: '*'
logging:
enabled: true
capture_payload:
request_response: false
$$;
What happens in this code:
DESCRIBEdisplays the active YAML specification.ALTERreplaces the specification entirely, exposing all available models ('*') and enabling trace logging without storing prompt/response payloads.
Limitations and requirements
- The feature is in Open Public Preview across AWS, Azure, and GCP commercial regions (excluding AWS Europe/Spain).
- Users require
USAGEprivilege on the gateway to submit inference requests. - Gateway permissions do not bypass model access controls; users still need permissions to underlying models, such as the
SNOWFLAKE.CORTEX_USERdatabase role. - Updating specifications via
ALTERreplaces the entire definition rather than patching individual fields.
- snowflake cortex
- cortex ai gateway
- ai governance
- llm observability
Cortex AI Gateway on Azure, Google Cloud, and AWS New Zealand
Resumo e explicação gerados por IA a partir da release note oficial. Em caso de dúvida, a documentação oficial da plataforma prevalece.
Relacionadas
- SnowflakeGA
Upgrade direto para Apache Iceberg v3 e expressões padrão
Agora você pode atualizar tabelas Apache Iceberg da versão 2 para a 3 sem recriá-las, mantendo permissões e referências, além de usar expressões SQL como valores padrão em tabelas v3 gerenciadas.
apache iceberg · iceberg v3 - SnowflakeGA
Espelhamento de dados do Snowflake Postgres chega em GA
Replique dados do Snowflake Postgres diretamente para o Snowflake quase em tempo real, sem conectores externos ou pipelines ETL, com suporte a tabelas nativas ou Iceberg.
snowflake postgres · data mirroring - SnowflakeGA
Snowflake Data Clean Rooms recebe versão 18.9 da API
A versão 18.9 da API do Snowflake Data Clean Rooms traz correções de bugs e melhorias de desempenho, com suporte a atualização automática ou manual.
snowflake · data clean rooms