Amplify LLM Model Upgrade: October 2026

On October 16, Bullhorn is upgrading the language model that powers Amplify to a newer, more current model. The switch will happen automatically, so there's nothing you need to do to receive it. Because a new model can read the same prompt slightly differently, we recommend testing your prompts and making any adjustments before the change goes live.

What to expect

You may notice slight differences in responses, but they should stay close to what you're used to. Format, level of detail, or ranking can shift even when nothing is wrong with your prompt. If that happens, a few small adjustments usually bring things back: say what output format you want, list your priorities, and any rules the model should follow.

Testing the new model

Amplify Admins can test existing prompts on the new model now:

  1. Go to Menu > Admin > Amplify Admin.

  2. Click the Prompts tab.

  3. Click the edit icon next to the prompt you want to test.

  4. In the right-hand panel, turn on Use New LLMClosed An AI language model (or LLM) is a system trained on large amounts of text that can read, interpret, and generate human-sounding language. Amplify uses one to read your Bullhorn records and produce output based on the prompts you configure. It does not search the internet or access outside data on is own. It only works with what you give it. Model.

  5. Click Update to generate a response with the new model.

Compare the result with what you get today, and adjust the wording or other parameters if you'd like to fine-tune it. We suggest starting with your company-wide prompts and the ones your teams rely on most.

The Use New LLM Model toggle applies only to the prompt you're testing. You'll need to turn it on each time you test a different prompt. Use New LLM Model doesn't switch on the new model anywhere else in Bullhorn or for any other users.

FAQ

Do I need to do anything to get the new model?

No, the upgrade happens automatically on October 16. We do recommend that admins test company-wide prompts beforehand, so you can be confident they still give you the results you want.

Can non-admin users test the new model?

No. The test option is available only in Prompt Studio within Amplify Admin, so it's limited to admins who can manage prompts.

What if a prompt doesn't work as well on the new model?

Small adjustments, such as being explicit about output format, priorities, and any rules the model should follow, usually restore the results you expect. Our Amplify Prompting Best Practices article has more guidance on building effective prompts.