Learn about support for new large language models (LLMs) for embedded features, such as Prompt Builder. Also, stay up-to-date with model deprecations and reroute dates.
- Einstein Models API Access Restricted for Experience Cloud Guest Users
Unauthenticated Experience Cloud guest users can no longer access the Einstein Models API. Requests from these users now return an HTTP 403 response. - GPT-5.6 Is Generally Available
The GPT-5.6 Luna, Terra, and Sol models are ready for use in production with a rate limit of 5,000 requests per minute. Use these models for certain functionality, such as in a prompt template. - GPT-5.5 Is Generally Available
This model was previously available as a beta model. It’s now ready for use in production. You can use this model in any supported feature, such as in a prompt template. - GPT-5.4 Mini Is Generally Available
This model was previously available as a beta model. It’s now ready for use in production. You can use this model in any supported feature, such as in a prompt template. - Claude Sonnet 5 Is Generally Available
This model is ready for use in production with a rate limit of 1,000 requests per minute. Use this model for certain functionality, such as in a prompt template. - Use Claude Opus 4.8 on the Einstein Platform
Claude Opus 4.8 is generally available. Use this model in any supported feature, such as in a prompt template. - Claude Opus 4.7 Is Generally Available
This model was previously available as a beta model. It’s now ready for use in production. You can use this model in any supported feature, such as in a prompt template. - Gemini 3.1 Flash Lite Is Generally Available
This model was previously available as a beta model. It’s now ready for use in production. You can use this model in any supported feature, such as in a prompt template. - Use Gemini 3.5 Flash on the Einstein Platform
Gemini 3.5 Flash is generally available. Use this model in any supported feature, such as in a prompt template. - Increased Limits for Large Language Model Generation Requests
The maximum requests per minute (RPM) for generation requests per org has increased from 1,000 to 2,000. This change means that an org can make up to 2,000 requests per minute across all the available models. Due to the available capacity from model providers, the RPM limit can vary for each individual model. The RPM and tokens per minute (TPM) limits have increased for several models. - Use Nemotron 3 Super 120B on the Einstein Platform (Beta)
We added support for Nemotron 3 Super 120B as a beta model. To use a beta model, you must first turn on beta generative AI models in the Einstein Setup page. After you turn on beta generative AI models, you can use this model in any supported feature, such as in a prompt template. - Claude Opus 4.6 Is Generally Available
This model was previously available as a beta model. It’s now ready for use in production. You can use this model in any supported feature, such as in a prompt template. - Prepare for Claude Sonnet 4 Reroute Date
Claude Sonnet 4 model requests will be rerouted to Claude Sonnet 4.6 on May 26, 2026 because Claude Sonnet 4 is in the Legacy state on Amazon Bedrock. We recommend that you test your prompts and apps with the new model in Prompt Builder and AI Models as soon as possible, because expected responses can change.