Add external models

This guide covers how to add and manage external AI models in AI Studio. After configuring external models, your organization can use models from providers like AWS Bedrock and OpenRouter alongside Writer’s Palmyra models when building agents.

External models are available on enterprise plans. Org admins, IT admins, and users with AI Studio full access roles can add and manage models. For billing, see Understand external model billing.

AI Studio supports text generation and embedding models from external providers. Video, audio, and image generation models are not supported.

Choose how to add models

Beyond Writer’s Palmyra models, you can add third-party models to your organization in two ways:

WRITER-managed models: Select from an array of third-party models that WRITER hosts and manages, such as models from Anthropic, DeepSeek, and Google. You don’t bring your own credentials—WRITER meters usage and bills you directly at the rates shown in the WRITER catalog. Enable the models you want from the WRITER catalog on the Models page.
Bring your own model (BYOM) through bring your own key (BYOK): Connect models from a provider like AWS Bedrock or OpenRouter using your own provider credentials. Your provider bills usage directly, and WRITER adds a 10% pass-through fee, paid via credits.

WRITER-managed third-party models are available in WRITER Agent only. They are not yet available through the API or SDK. BYOM models connected with your own credentials are available across WRITER Agent and the API.

The rest of this guide covers adding and managing BYOM models with your own credentials.

How external models work

The Models page in AI Studio provides a unified governance layer for managing AI models across your organization. You can add models from external providers and control which teams have access to use them when building agents.

The Models page also displays Writer’s Palmyra models. Palmyra models are always available to all teams and don’t require configuration. Select any Palmyra model to view its details.

External model requests flow through the following steps:

Credential configuration: Administrators add provider credentials (API keys or IAM roles) to AI Studio
Model selection: Administrators choose which models from the provider to enable
Access control: Administrators assign model access to all teams or specific teams
Agent building: Developers with access can select the model when building agents
Request routing: When agents run, requests route through the configured provider credentials

Understand external model billing

External models use bring your own model (BYOM) billing. Your provider bills usage directly at its rate card, and WRITER adds a 10% pass-through fee on that usage, paid via credits. This differs from WRITER-managed models, which are metered entirely through WRITER. For rates and details, see External models (BYOM) on the pricing page. To track spend, see the Monitor costs section on each provider page, such as AWS Bedrock or OpenRouter.

Available providers

AI Studio supports external models from the following providers. Select a provider to view detailed configuration instructions.

Provider	Status	Auth type	Provider docs
AWS Bedrock	Available	Access keys or IAM Role ARN	View docs
OpenRouter	Available	API key	View docs

AWS Bedrock models are currently supported in the following regions: us-east-1, us-west-1, us-west-2, and eu-west-1. Additional regions may be added in future releases.

Add an external model

Add external models in AI Studio under Models & Guardrails > Models.

Add a model in AI Studio

For reusability and easier rotation, create named credentials before adding a model. You can also enter credentials inline during the Add Model flow (see Enter credentials directly).

Navigate to Models & Guardrails > Models in AI Studio
Select + Add model
Choose your provider (for example, AWS Bedrock or OpenRouter)
Enter your credentials
Choose which models to enable from the available list
Configure team access (all teams or specific teams)
Select Add model to complete the setup

You can reuse credentials across multiple models. See Manage credentials for details.

Manage team access

Control which teams can use external models when building agents.

Configure model availability

When adding a model, you can set access to:

All teams: The model is immediately available to everyone with builder access
Specific teams: Restrict the model to selected teams

Update team access

To view or update which teams can access a model:

Navigate to Models & Guardrails > Models in AI Studio
View the current team access in the Team Access column
Select the menu icon and choose Edit to update team access

Credentials are configured at the model level, not the team level. Team access controls who can use the model, but all authorized users share the same underlying provider credentials. Team members with access can use the models but cannot view the credentials.

Manage credentials

AI Studio stores provider credentials as named credential sets that you can reuse across multiple models. You can manage credentials from the dedicated LLM Credentials page or create them when adding a model.

Create credentials from the LLM Credentials page

To create credentials before adding models:

Navigate to Models & Guardrails > LLM Credentials in AI Studio
Select Add credentials
Enter a Credential name (for example, dev-aws-credentials or prod-openrouter)
Select the Provider (for example, Bedrock or OpenRouter)
Enter the provider-specific authentication details
Select Save

Enter credentials directly when adding a model

When adding a model, you can enter credentials directly in the Add Model form instead of selecting existing credentials. Credentials entered this way are stored with the model but are not saved as a named credential set for reuse with other models. To create reusable credentials that you can share across multiple models, use the LLM Credentials page instead.

Reuse credentials across models

When adding additional models from the same provider:

Select your existing credentials from the Credentials name dropdown
The stored authentication details are automatically applied
You don’t need to re-enter access keys or other sensitive values

This approach simplifies management when you have multiple models from the same provider—update credentials in one place and all associated models use the updated values.

Update credentials

How you update expired or rotated credentials depends on how you originally configured them: Named credentials (created on the LLM Credentials page):

Navigate to Models & Guardrails > LLM Credentials in AI Studio
Locate the credential in the list
Select the menu icon and choose Edit credentials
Update the authentication details and save

All models using this credential automatically use the updated values. Inline credentials (entered directly when adding a model):

Navigate to Models & Guardrails > Models in AI Studio
Locate the model in the list
Select the menu icon and choose Edit
Update the credential values and save

For easier credential management, use named credentials from the LLM Credentials page. Named credentials can be updated in one place and automatically apply to all models using them.

Delete credentials

To delete a named credential:

Navigate to Models & Guardrails > LLM Credentials in AI Studio
Locate the credential in the list
Select the menu icon and choose Delete

Before deleting credentials, verify that no models are using them. Models using deleted credentials will fail to authenticate with the provider.

Security best practices

Follow these practices when managing provider credentials:

Use dedicated service accounts rather than personal credentials
Rotate credentials on a regular schedule (for example, every 90 days)
Use descriptive credential names that indicate environment or purpose (for example, prod-bedrock-us-east)
For AWS, prefer IAM Role ARN over access keys when possible

Edit a model

To update an external model’s configuration, including credentials and team access:

Navigate to Models & Guardrails > Models in AI Studio
Locate the model in the list
Select the menu icon and choose Edit
Update the configuration and save

Delete a model

To remove an external model from AI Studio:

Navigate to Models & Guardrails > Models in AI Studio
Locate the model in the list
Select the menu icon and choose Delete

Before deleting a model, verify that no active agents depend on it. Agents that reference a deleted model will return a “model not found” error when they run. You’ll need to update the agent to use a different model before it can run successfully.

Use external models

Once you add an external model, it’s available to use the same way as Palmyra models.

Agent Builder and no-code apps

In Agent Builder and no-code chat apps, external models appear in the Model dropdown alongside Palmyra models. Select any model your team has access to.

API usage

External models use the same Writer API as Palmyra models. Use the List models endpoint to see all available models and their IDs, then pass the model ID to any endpoint that supports the model parameter.

curl https://api.writer.com/v1/chat/completions \
  -H "Authorization: Bearer $WRITER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic.claude-3-sonnet-20240229-v1:0",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from writerai import Writer

client = Writer()

response = client.chat.chat(
    model="anthropic.claude-3-sonnet-20240229-v1:0",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

import Writer from "writer-sdk";

const client = new Writer();

const response = await client.chat.chat({
  model: "anthropic.claude-3-sonnet-20240229-v1:0",
  messages: [{ role: "user", content: "Hello!" }],
});
console.log(response.choices[0].message.content);

Next steps

Configure AWS Bedrock: Set up AWS credentials and add Bedrock models
Configure OpenRouter: Set up an OpenRouter API key and add OpenRouter models
Choose a model: Compare Palmyra models with external provider models
Palmyra models: Explore Writer’s Palmyra model capabilities

Getting started

Core concepts

Agent Builder v2

Monitor and manage

Models and pricing

Connect to your data

Chat completions

No-code agents

Knowledge Graphs

Tool calling

Additional capabilities

Integrations

Security and compliance

Resources

Choose how to add models

How external models work

Understand external model billing

Available providers

Add an external model

Add a model in AI Studio

Manage team access

Configure model availability

Update team access

Manage credentials

Create credentials from the LLM Credentials page

Enter credentials directly when adding a model

Reuse credentials across models

Update credentials

Delete credentials

Security best practices

Edit a model

Delete a model

Use external models

Agent Builder and no-code apps

API usage

Next steps

​Choose how to add models

​How external models work

​Understand external model billing

​Available providers

​Add an external model

​Add a model in AI Studio

​Manage team access

​Configure model availability

​Update team access

​Manage credentials

​Create credentials from the LLM Credentials page

​Enter credentials directly when adding a model

​Reuse credentials across models

​Update credentials

​Delete credentials

​Security best practices

​Edit a model

​Delete a model

​Use external models

​Agent Builder and no-code apps

​API usage

​Next steps

Choose how to add models

How external models work

Understand external model billing

Available providers

Add an external model

Add a model in AI Studio

Manage team access

Configure model availability

Update team access

Manage credentials

Create credentials from the LLM Credentials page

Enter credentials directly when adding a model

Reuse credentials across models

Update credentials

Delete credentials

Security best practices

Edit a model

Delete a model

Use external models

Agent Builder and no-code apps

API usage

Next steps