Bring your own AI model, and put your brand on Gydr

Every business that puts an AI assistant in front of its customers eventually asks two questions. Whose AI model is doing the answering? And whose name is on the product?

Until this release, Gydr answered both for you. With Gydr 1.31, you can answer them yourself. You can bring your own AI model, connecting your own provider and choosing the model behind each chatbot. You can rebrand the Console your team works in. And you can test your chatbot's personality before your customers ever hear it.

The model is a component, not the product

Some businesses already have an agreement with a model provider. Some have data commitments that name one, and some simply prefer one. Bring Your Own Model lets you connect OpenAI, Anthropic, Gemini, DeepSeek, Qwen, Mistral, Moonshot, OpenRouter or Groq from a new Import Models page in Settings. You add and check each connection there, then pick a model per widget. Conversations on a widget that runs your own model bill to your provider account, not to your Gydr credits.

This reflects how we think about models in general. A model is a powerful component, and the right one changes every few months. What makes an assistant trustworthy is everything around the model: the knowledge it is allowed to use, the instructions it follows, and the checks on what it says before a customer reads it. Those stay the same whichever model you choose. Swapping the engine should never mean rebuilding the car.

A Console that carries your name

Agencies, resellers and larger teams often run Gydr for someone else, or for many someones. White Label replaces the Gydr logo and accent colour across the Console's sidebar and top bar with your own, for every member of your team. Your clients see your product.

A personality you can hear before you ship it

Tone is part of a brand. A chatbot for a law firm should not sound like one for a surf shop. You can choose from eight ready-made personalities or describe your own style. Then try it in a live sandbox and read how it responds before you save. You hear it first, not your customers.

The rest of the release

We rebuilt the Console from the ground up. It is mobile-first, faster, and available in English, Malay and Chinese. Usage History now breaks credits into clear categories: chat, knowledge retrieval, tools, website crawling and file attachments. You can see where your credits go instead of guessing.

Two features arrived in preview. Visitors can attach files while they talk to one of your agents, and the files reach your help desk with the transcript. Credit alert emails warn the people you choose before your credits run out.

One change is invisible but matters most. Each widget now draws only on the knowledge bases and skills you assign to it, and that rule holds at every step, from search to answer. A chatbot for one brand can never quote the catalogue of another, even inside the same account.

Frequently asked questions

Which AI model providers can I bring to Gydr?

At launch in Gydr 1.31: OpenAI, Anthropic, Gemini, DeepSeek, Qwen, Mistral, Moonshot, OpenRouter and Groq. Requesty joined the list in Gydr 2.6.

Who pays for conversations that run on my own model?

Your model provider does. Conversations on a widget that runs your own model bill to your provider account, not to your Gydr credits.

What does White Label change in the Console?

It replaces the Gydr logo and accent colour across the Console for every member of your team. Later releases extended White Label to your own domains and to the emails your users receive.

The full list of changes is in the 1.31.0 release notes.