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.
- Prepare for Claude 3.7 Sonnet Reroute Date
Claude 3.7 Sonnet is deprecated on Amazon Bedrock (Anthropic). Claude 3.7 Sonnet model requests will be rerouted to Claude Sonnet 4.5 on February 26, 2026. We recommend that you test your prompts and apps with the new model in Prompt Builder and Einstein Studio as soon as possible, because expected responses can change. - Prepare for Claude 3 Haiku Reroute Date
Claude 3 Haiku is deprecated on Amazon Bedrock (Anthropic). Claude 3 Haiku model requests will be rerouted to Claude Haiku 4.5 on February 26, 2026. We recommend that you test your prompts and apps with the new model in Prompt Builder and Einstein Studio as soon as possible, because expected responses can change. - Prepare for Gemini 2.0 Flash and 2.0 Flash Lite Reroute Date
Gemini 2.0 Flash and Gemini 2.0 Flash Lite are deprecated on Vertex AI (Google). Gemini 2.0 Flash and Gemini 2.0 Flash Lite model requests will be rerouted to Gemini 2.5 Flash and Gemini 2.5 Flash Lite respectively on February 20, 2026. We recommend that you test your prompts and apps with the new models in Prompt Builder and Einstein Studio as soon as possible, because expected responses can change. - Use NVIDIA Nemotron 3 Nano 30B on the Einstein Platform (Beta)
We added support for NVIDIA Nemotron 3 Nano 30B 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. - Use Gemini 3.1 Pro on the Einstein Platform (Beta)
We added support for Gemini 3.1 Pro 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. - Gemini 3 Pro (Beta) Is Retiring
This preview model from Google is retiring. Since Gemini 3 Pro was in Beta and will never be generally available, the model is retiring and will not be automatically rerouted. We recommend that you migrate your applications away from Gemini 3 Pro (Beta) as soon as possible. During migration, update and test each part of your application with the replacement model that we recommend, which is Gemini 3.1 Pro (Beta). - GPT 5.1, GPT 5.2, and Claude Opus 4.5 Are Generally Available
These models were previously available as beta models. They're now ready for use in production. You can use these models in any supported feature, such as in a prompt template. - Use Gemini 3.1 Flash Lite on the Einstein Platform (Beta)
We added support for Gemini 3.1 Flash Lite 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. - Use GPT 5.4 on the Einstein Platform (Beta)
We added support for GPT 5.4 as a beta model. This geo-aware model uses OpenAI or Azure OpenAI to service LLM requests. The same underlying model is used by OpenAI and Azure OpenAI. 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. - Use Claude Sonnet 4.6 on the Einstein Platform
Claude Sonnet 4.6 is generally available. Use this model in any supported feature, such as in a prompt template. - Einstein Studio Is Now AI Models
Einstein Studio is now called AI Models in Data 360. During this transition, you may see a mix of the new and previous terms in our documentation and UI. - Understand the Tokens Per Minute Model Limit
For service protection, there's a new tokens per minute (TPM) limit per org for each model. We measure an org's model usage against the sum of input and output tokens in one minute. If the TPM limit for a model is exceeded, an error is returned. - Use GPT 5.4 on the Einstein Platform
GPT 5.4 is generally available. It was available as a beta model earlier in Spring ’26. This geo-aware model uses OpenAI or Azure OpenAI to service LLM requests. The same underlying model is used by OpenAI and Azure OpenAI. Use this model in any supported feature, such as in a prompt template. - Use Claude Opus 4.6 on the Einstein Platform (Beta)
We added support for Claude Opus 4.6 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. - Use Claude Opus 4.7 on the Einstein Platform (Beta)
We added support for Claude Opus 4.7 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. - Use GPT 5.4 Mini on the Einstein Platform (Beta)
We added support for GPT 5.4 Mini as a beta model. This geo-aware model uses OpenAI or Azure OpenAI to service LLM requests. The same underlying model is used by OpenAI and Azure OpenAI. 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. - Use GPT 5.5 on the Einstein Platform (Beta)
We added support for GPT 5.5 as a beta model. This geo-aware model uses OpenAI or Azure OpenAI to service LLM requests. The same underlying model is used by OpenAI and Azure OpenAI. 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.