Getting Started

Set up AI assistance in Positron: configure a language model provider such as Posit AI, Anthropic, or GitHub Copilot, then open Posit Assistant.

Posit Assistant is Positron’s new AI chat experience, available as a preview feature in Positron 2026.04.0 and later. To get set up, see the Posit Assistant documentation.

Positron Assistant is being deprecated

Positron Assistant is the old chat experience and is being superseded by Posit Assistant. It will be removed in a future release, so we recommend using Posit Assistant instead. Positron Assistant is still available for now, and the rest of this guide covers it for anyone who still needs it in the meantime.

Both assistants connect to the same language model providers, and you configure them the same way. The following providers are supported:

Provider	Chat	Code Completions	Authentication
Posit AI			OAuth
Anthropic			API Key
GitHub Copilot (Preview)			GitHub OAuth
Amazon Bedrock			AWS CLI or Posit Workbench Managed Credentials
Snowflake Cortex			API Key or Posit Workbench Managed Credentials
Microsoft Foundry (Preview)			API Key or Posit Workbench Managed Credentials
OpenAI			API Key
Google Vertex AI (Experimental)			Google Cloud CLI or Service Account Credentials
Custom Provider (Experimental)			API Key

We’re actively expanding provider support and welcome your feedback and issue reports!

Please ensure you consult your provider’s privacy policy for information on the data they collect and how it is used. For reference links, see the Provider Privacy Information guide.

For information on what information is collected by Posit, review Posit’s privacy policy.

Step 1: Configure a language model provider

Posit AI, Anthropic, GitHub Copilot, Amazon Bedrock, Snowflake Cortex, and OpenAI are enabled by default but are not active until you sign in. Other providers are in preview or experimental status and must be enabled through settings.

Important

Posit AI requires an account. Visit posit.ai to learn more and create one.

Posit AI is a hosted language model service from Posit. It provides both chat models and code completions in the form of Next Edit Suggestions (NES).

Add Posit AI as a language model provider

Run the command Authentication: Configure Language Model Providers
Select Posit AI as the model provider.
Select the Sign in button to initiate Posit AI’s OAuth authentication flow.
- Complete the authentication flow in your browser, and return to Positron when finished.

Important

Authentication requires a Claude Console account with API access. See Anthropic’s Claude API and Console documentation for more information.

Authentication does not support logging into Anthropic with the Claude Pro Plan, Max Plan, or other Claude subscription plans.

How to get an Anthropic API key

To use Anthropic’s Claude models in Positron, you need to bring your own API key (BYOK). To obtain an API key from Anthropic:

Log in to or create an account for Anthropic’s Claude Console.
Navigate to the API keys management page.
Select the Create Key button.
Fill out any required information and select Add to generate your API key.
Copy and save the API key to a password manager or another secure location.

Add Anthropic as a language model provider

Run the command Authentication: Configure Language Model Providers
Select Anthropic as the model provider.
Paste your Claude Console API key into the input box and select Sign in.

Note

Alternatively, set the ANTHROPIC_API_KEY environment variable to authenticate with Anthropic in Positron.

Important

GitHub Copilot requires a GitHub account with Copilot enabled.

How to get GitHub Copilot access

GitHub Copilot is a proprietary tool from GitHub. If you want to use GitHub Copilot, you need a subscription for GitHub Copilot in your personal GitHub account or to be assigned a seat by an organization with a subscription for GitHub Copilot for Business.

Students and faculty can use GitHub Copilot for free as part of the GitHub Education program. For more information, see the GitHub Education page.

Add GitHub Copilot as a language model provider

Run the command Authentication: Configure Language Model Providers
Select GitHub Copilot as the model provider.
Select the Sign in button to initiate GitHub’s OAuth authentication flow.
- Complete the authentication flow in your browser, and return to Positron when finished.

Note

You can also sign in to GitHub via the Accounts icon in the Activity Bar. To sign out, use the Accounts icon or run the Accounts: Manage Accounts command.

Note that the Accounts section manages your GitHub authentication for all extensions that use GitHub, not just AI features.

Important

To use Amazon Bedrock in Positron Desktop, you must:

Have an AWS account with Amazon Bedrock access
Have access to foundation models through the AWS Bedrock console (request access to the foundation models)
Sign in to your AWS account using the AWS CLI

(Optional) Configure AWS region and profile

Positron reads AWS_REGION from your environment automatically. You only need to set it here if your shell does not already have AWS_REGION configured and your Bedrock-enabled account is not in us-east-1 (the default). Use the standard AWS region identifier format (for example, us-east-1, eu-west-1, ap-southeast-1). The Amazon Bedrock endpoints and quotas reference lists the region identifier for each location. You must also ensure the models you want to use are available in that region.

Positron uses your default AWS CLI profile when no profile is specified. Set AWS_PROFILE if the profile you use for Bedrock is not named default — for example, when you have a dedicated Bedrock role separate from your general development credentials, or when your Bedrock access lives in a different AWS account. To see your available profiles, run aws configure list-profiles in a terminal.

Note

Do not set AWS_PROFILE if you use Posit Workbench managed credentials or environment variable credentials (AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY). Both of those paths bypass named profile resolution, and setting a profile name in those cases causes credential resolution to fail.

To set either value, add them to the positron.assistant.providerVariables.bedrock setting. These override any matching environment variables already set on your system for Positron only; they do not affect your terminal or other tools. Changes require a restart to take effect.

(Optional) Configure inference profile region

Positron uses cross-region inference profiles to route requests across multiple AWS regions for higher availability and throughput. Positron derives the inference profile region automatically from AWS_REGION by taking its geographic prefix:

`AWS_REGION`	Derived inference profile region
`us-east-1`, `us-west-2`	`us`
`eu-west-1`, `eu-central-1`	`eu`
`ap-southeast-1`, `ap-northeast-1`	`apac`

Most users do not need to change this. Use the positron.assistant.bedrock.inferenceProfileRegion setting to override the derived value if:

Your AWS account has Bedrock model access enabled in a different geographic pool than your AWS_REGION implies — for example, you connect through us-east-1 but want to route inference through eu profiles.
You want to use a global inference profile, which Positron cannot derive automatically from any region.

Valid values are us, eu, apac, and global. This setting requires a restart to take effect.

Note

If no inference profile exists for your preferred region, Positron falls back to the first available inference profile for that model. If a model has no inference profile at all, it does not appear in the model list.

Add Amazon Bedrock as a language model provider

Run the command Authentication: Configure Language Model Providers
Select Amazon Bedrock as the model provider.
Select Sign in so Positron can verify your AWS CLI authentication.

Important

To use Snowflake Cortex in Positron Desktop, you need a Snowflake account with Cortex access, as well as a Snowflake account identifier and programmatic access token (PAT).

Add Snowflake Cortex as a language model provider

Add your SNOWFLAKE_ACCOUNT ID to the positron.assistant.providerVariables.snowflake setting
Run the command Authentication: Configure Language Model Providers
Select Snowflake Cortex as the model provider.
Paste your Snowflake Cortex PAT into the API key box, then select Sign in.

Important

To use Microsoft Foundry in Positron Desktop, you must:

Have a Microsoft Foundry resource with one or more models deployed
Have the endpoint URL and API key for your Foundry resource

Foundry does not support automatic model discovery. Positron defaults to the model-router identifier, but you can configure specific models. See Configure custom model overrides below.

Enable Microsoft Foundry as a provider

Microsoft Foundry provider support is in preview. Opt in to the positron.assistant.provider.msFoundry.enable setting to enable it.

Add Microsoft Foundry as a language model provider

Run the command Authentication: Configure Language Model Providers
Select Microsoft Foundry as the model provider.
Enter your Foundry endpoint URL in the Base URL field and your API key in the API Key field, then select Sign in.

Note

Positron accepts deployment-based Azure OpenAI URLs (for example, https://<resource>.openai.azure.com/openai/deployments/<name>/chat/completions?api-version=...) and automatically normalizes them to the supported v1 format. A message displays when this conversion occurs.

Configure custom model overrides

To use specific models instead of model-router, configure the positron.assistant.models.overrides.msFoundry setting:

"positron.assistant.models.overrides.msFoundry": [
  {
    "name": "gpt-4.1",
    "identifier": "gpt-4.1"
  },
  {
    "name": "gpt-5.3-chat",
    "identifier": "gpt-5.3-chat"
  }
]

Use the model name and deployment identifier from your Foundry resource for each entry.

Add OpenAI as a language model provider

Run the command Authentication: Configure Language Model Providers
Select OpenAI as the model provider.
Paste your OpenAI API key into the input box and select Sign in.

Note

Use the OpenAI provider to connect to a provider that implements the OpenAI Responses API, by replacing the base URL with your compatible custom endpoint.

Important

To use Google Vertex AI in Positron Desktop, you must:

Have a Google Cloud project with the Vertex AI API enabled
Have access to the foundation models you want to use through Vertex AI Model Garden
Authenticate with Google Cloud using one of the supported credential flows below

Install the Google Cloud CLI (gcloud)
Run gcloud auth application-default login and complete the browser sign-in flow

Set the following environment variables before launching Positron:

GOOGLE_CLIENT_EMAIL: the service account’s client email
GOOGLE_PRIVATE_KEY: the service account’s private key

Enable Google Vertex AI as a provider

Google Vertex AI provider support is experimental. Opt in to the assistant.provider.googleVertex.enabled setting to enable it.

Add Google Vertex AI as a language model provider

Set your Google Cloud project ID and location, using either environment variables or settings:
- Environment variables: Set GOOGLE_VERTEX_PROJECT and GOOGLE_VERTEX_LOCATION before launching Positron.
- Settings: Add GOOGLE_VERTEX_PROJECT and GOOGLE_VERTEX_LOCATION entries to the authentication.googleVertex.credentials setting. Settings take precedence over environment variables.
Run the command Authentication: Configure Language Model Providers
Select Google Vertex AI as the model provider.
Select Sign in to authenticate with Google Vertex AI.

Important

Positron’s custom provider support is intended for use with any OpenAI-compatible API endpoint that uses the /v1/chat/completions endpoint for chat.

We don’t recommend using a local model as a custom provider. Read more about why local models are not there (yet) on Posit’s blog.

Enable Custom Provider as a provider

Custom provider support is experimental. Opt in to the positron.assistant.provider.customProvider.enable setting to enable it.

Add Custom Provider as a language model provider

Run the command Authentication: Configure Language Model Providers
Select Custom Provider as the model provider.
Enter your API key and base URL into the input boxes and select Sign in.

Note

Some OpenAI-compatible providers may not implement the /models endpoint, which Positron uses to list available models. If this is the case for your provider, you can manually configure a model listing using the positron.assistant.models.overrides.customProvider setting.

Step 2: Open your assistant

With a language model provider configured, open the assistant you want to use.

Posit Assistant (recommended)

Posit Assistant is pre-installed in recent versions of Positron. Run the View: Show Posit Assistant command in the Command Palette to open the chat panel and start a conversation.

For the full setup guide and feature documentation, see the Posit Assistant documentation.

Note

To use GitHub Copilot with Posit Assistant, you must also enable Positron Assistant by turning on the old positron.assistant.enable setting.

At this time, Posit Assistant relies on Positron Assistant for GitHub Copilot authentication. Once Positron Assistant is deprecated, Posit Assistant will no longer require this extra step for GitHub Copilot support.

Positron Assistant (being deprecated)

Positron Assistant is being deprecated, and we recommend Posit Assistant instead. If you still need Positron Assistant:

Enable the setting positron.assistant.enable.
Restart Positron or run the Developer: Reload Window command in the Command Palette.
Select the chat robot icon in the sidebar, or run the command Chat: Open Chat in the Command Palette to open the chat:

Assistant Sidebar Icon
Chat with Assistant by typing your question or request in the chat input box at the bottom of the chat pane, then pressing Enter or the send button.

To learn more about Positron Assistant’s features, check out the following guides:

Chat pane

A dedicated AI chat pane for coding, troubleshooting, and planning.

Inline chat

In-context AI assistance within your code Editor, Notebook, or Terminal.

Code completions

Receive code suggestions directly in your Editor or Notebook.