Improved chat experience: voice input, model recommendations and more

The 9brains chat has been thoroughly reworked. The result: a more modern, more reliable chat experience with several new features.

Voice input

You can now dictate your messages by voice. Click the microphone icon in the input field, speak your message, and the transcribed text appears in the input field automatically. Speech recognition works in all common browsers.

Intelligent model selector

Model selection has been simplified. Instead of a long list of every available model, you now see three curated recommendations:

• Premium: the most capable model for complex analyses and coding
• Allround: our recommendation for everyday work with the best balance of quality and cost
• Budget: ultra-economical for simple tasks at solid quality

Via the “Show all models” button you still have access to every available model. In addition, each model now shows how much an average request costs, so you can decide based on the facts.

EU-only mode

Administrators can now configure in the model settings that only models with EU data processing are selectable. A single button deactivates every non-EU model for the entire workspace, ideal for companies with strict data protection requirements.

Model intelligence and cost for admins

In the model settings, administrators now see two additional metrics: the average cost per request and an IQ value that ranks the model’s performance. This makes it easier to decide which models to enable for the workspace. An interactive overview of all models with cost and intelligence is available under AI models.

Edit, regenerate and copy messages

The chat now offers three new actions per message:

• Regenerate: have the AI produce a new response when the result does not fit
• Edit: adjust your own message afterwards and receive a new response
• Copy: copy any response as formatted Markdown text to the clipboard

Message feedback

You can now react to every AI response with thumbs up or thumbs down. Your feedback helps us continuously improve the quality of responses.

Visible reasoning and tool usage

For models that support a reasoning process, you can now trace how the AI arrived at its answer. The use of tools (such as web search or document analysis) is also displayed transparently in the chat.

Improved scrolling

Scrolling behavior in the chat has been optimized. While you read a response, the view no longer scrolls away automatically. New content only appears once you scroll to the end of the conversation.

Faster search in the chat list

Search in the sidebar now automatically searches older chats on the server as well, without you having to click an extra button. Results from your local chats appear instantly, results from the server are loaded in seamlessly.