Anthropic
Snowflake’s partnership with Anthropic provides customers with launch day access to Claude models, along with commercial advantages like buying Snowflake via Claude Marketplace to streamline procurement and reduce cost.
AI Model Providers
Choose the AI model that fits each task. Models run alongside the Snowflake data, semantic definitions and access controls you already govern, providing the enterprise context needed to securely produce accurate, verifiable answers.
The Snowflake advantage
Frontier models get better when they understand your business
Getting more from AI models starts with giving them the right context. Snowflake is the governed context layer between your data and the AI models you choose, with Cortex AI Gateway* controlling access, routing and spend centrally. Match the right model to each task and switch model providers freely, without rebuilding your context or governance layer.
Choose your model and your cloud
Run models from Anthropic, DeepSeek*, Google, Meta, Mistral AI, OpenAI and SpaceXAI across AWS, Azure and Google Cloud from a single connection.
Give full context for more accurate results
Models read the semantic definitions, data relationships and access policies in Snowflake, so answers reflect your organization's own intelligence.
Control model spend from one control plane
Set budgets by team or cost center, apply per-user credit quotas and see token-level usage across every model and agent in one place with Cortex AI Gateway*.
Run models where your data lives
Inference happens inside Snowflake's security and governance perimeter, under the access controls you already enforce.
How it works
Get started in three steps
Connect your data
Your Snowflake data and semantic definitions are already in place. AI models query that foundation directly, and nothing needs data copying, staging or manual re-permissioning.
Choose your model
Select from 20+ AI models in Snowflake Cortex AI, or let dynamic model routing in Cortex AI Gateway* automatically match each query to the best-fit model based on your cost, quality and speed targets.
Build and deploy
Use Snowflake Cortex AI to build, test and deploy AI applications and agents, with spend monitoring, access controls and full audit trails from day one.
AI Model Partners
Frontier intelligencein Snowflake Cortex AI
Leading AI models from the world's top labs, running inside Snowflake with your business context and governance capabilities built in.
A growing ecosystem of model partners
In the news
Latest from Snowflake
AI Model Providers
Frequently Asked Questions
Common questions about model choice, governance, data protection and cost when using AI model providers with Snowflake.
Which AI model providers does Snowflake support?
Snowflake Cortex AI gives you access to models from Anthropic, DeepSeek*, Google, Meta, Mistral AI, OpenAI, SpaceXAI and beyond. The supported model catalog grows regularly, so your organization can adopt new providers as they emerge with no changes to your data infrastructure or governance policies.
Can I switch between AI model providers without migrating my data or rebuilding my workflow?
Yes. Snowflake acts as the governed context and orchestration layer between your data and any AI model. Switching providers, or running multiple models simultaneously, does not require changes to your data architecture, semantic definitions or access controls. All of that stays in Snowflake.
How do I choose the right AI model for each task?
You don't have to choose just one. Dynamic model routing in Cortex AI Gateway* selects a model for each query based on your priorities, balancing quality, cost and speed. A latency-sensitive task can route to a faster, lower-cost model, while a complex reasoning task routes to a frontier model. You can also specify a model directly whenever you want explicit control.
Does Snowflake use my data to train AI models?
No. Snowflake never uses your customer data to train models made available to other customers. AI models run inside Snowflake's security and governance perimeter, so your data and the models that use it stay within the same boundary you already govern.
How does running AI on Snowflake improve the accuracy of model outputs?
AI models produce better results when they have full enterprise context: business definitions, semantic relationships, data permissions and organizational hierarchies. Snowflake consolidates that context in a single governed layer, so every model querying Snowflake data benefits from it without requiring the model provider to rebuild your governance infrastructure.
Where do AI models actually run?
Inference runs within the Snowflake perimeter. You control the regional boundary for inference at the account level, from any supported region down to your home region only.
How does AI model usage pricing work on Snowflake?
AI usage through Snowflake Cortex AI draws on your existing Snowflake compute credits, with no separate per-seat pricing or additional procurement required. Organizations that have standardized on Snowflake can adopt new AI models without new vendor contracts or budget approvals.
How do I control AI costs in Snowflake?
Cortex AI Gateway* is the central control point for AI consumption. Set monthly budgets for a team or cost center, apply per-user credit quotas that block further usage when someone reaches their limit and track token-level usage by model, user and workload. Account usage views report AI spend alongside the rest of your Snowflake consumption, so finance and platform teams work from one system of record.
