Claude models are not available in Loop

​

Summary

Issue: Claude model names aren’t selectable in the Loop model picker, even with an Anthropic or AWS Bedrock provider configured. Cause: Loop runs only on its own supported models. Configuring an AI provider doesn’t widen that set, so no Claude model becomes selectable in Loop. Resolution: Select one of Loop’s supported models, served either by Braintrust’s built-in models or by one of your own compatible providers.

​

Resolution steps

​

Step 1: Select a supported model

Open Loop, open the model picker, and choose one of Loop’s supported models. New threads start on GPT-5.6 Sol at medium reasoning effort, which Braintrust recommends.

​

Step 2: Confirm something can serve it

The Built-in setting in the model picker decides who serves the model:

• On. Braintrust serves it, and usage draws down your organization’s monthly model credits. This needs no provider setup.
• Off. One of your organization’s own AI providers serves it, and inference is billed to that provider. Compatible provider types are OpenAI, Azure, Bedrock, OpenRouter, and Azure AI Gateway, and the provider you select must serve the model you chose.

See Models and providers for how Loop resolves the two.

​

Notes

• Adding a custom model to a provider doesn’t make that model selectable in Loop. Loop offers only its supported models.
• Bedrock and the Anthropic API are separate provider types. Configuring one does not enable the other.
• Loop uses organization-level AI providers only. A project-level provider does not make models available in Loop. See Configure an AI provider for Loop.
• Claude models remain available everywhere else they were before, including prompts, playgrounds, and LLM-as-a-judge scorers. This applies only to the model Loop itself runs on.