[AWS Bedrock] Sonnet 4.5. model identifier not showing up

[FabianAmherd]

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

When trying to run Sonnet 4.5. inference with my own AWS Bedrock instance of it, the model does not show up in the settings “models” list. Confirmed via awscli that I do indeed have access to the model in the specified region.

The models showing up in the “models” list do not seem to be dynamically fetched from the users access.

Also, whenever I restart Cursor IDE, the “Test Model” field of the “AWS Bedrock” section is automatically set back to 3.5. sonnet in the us region, even if defined differently before.

Steps to Reproduce

1. Open Cursor IDE
2. activate Amazon Bedrock API Key
3. Find sonnet 4.5. in model picker

Operating System

Linux

Current Cursor Version (Menu → About Cursor → Copy)

Version : 1.7.40
Validation : 1.99.3
Date : df79b2380cd32922cad03529b0dc0c946c311850
Electron : 2025-10-09T02:55:11.735Z
ElectronBuildId : 34.5.8
Chromium : undefined
Node.js : 132.0.6834.210
V8 : 20.19.1
Système d’exploitation : 13.2.152.41-electron.0

For AI issues: which model did you use?

Sonnet 4.5.

Does this stop you from using Cursor

Yes - Cursor is unusable

[deanrie]

Hey, thanks for the report. You’re right, the model list isn’t dynamically fetched from your AWS. Cursor has a predefined list of supported models for Bedrock. Claude 3.5 Sonnet should be available, but if you’re looking for the very latest version, it might not be in the list yet.

Regarding the Test Model reset, that looks like a bug that needs to be fixed. Settings should persist across restarts.

A couple things to try:

1. Make sure you’re using the correct Bedrock model identifier format, it should look like anthropic.claude-3-5-sonnet-20241022-v2:0.
2. Try entering the model ID manually in the custom model field if it doesn’t appear in the dropdown.

If the model still doesn’t work even with the correct identifier, this may need to be escalated, since the Bedrock integration might require an update to support the newest model versions.

[FabianAmherd]

The key was to add the correct inference profile under the custom model field.