T-Chat 4.0 brings enabled online models into multi-agent meetings alongside local engines. Each participant can use the resource that best fits the task.
One meeting, different capabilities
A legal review, a creative proposal and a technical verification do not share the same requirements. T-Chat personalities can meet around one objective while working with different models.
The method stays consistent: roles, directives and the final outcome remain organised in the same environment.
Local when data and continuity come first
Local models are the natural choice when documents, context and conversations must remain on the Mac, or when work needs to continue without relying on a connection.
Companion shows which engines are genuinely available and how many resources they are using.
Online when a specific capability matters
An online service can help when a particular model is a stronger fit, when a larger context window is needed or when the local machine should remain free for other work.
When an external provider is selected, the data required for the request leaves the local perimeter and follows that provider's terms, privacy policy and pricing.
OpenRouter: one key, more possibilities
The OpenRouter integration connects models enabled on the account to T-Chat through a single configuration. That active catalogue can also be used in meetings.
Convenience does not replace governance: the company chooses which models are available and where they may be used.
Real freedom means being able to change
A system tied to one provider inherits its outages, pricing and rules. A hybrid architecture keeps more options available.
That does not mean using everything at once. It means assigning each task to the resource that best matches data sensitivity, speed, cost and continuity.
An AI strategy should not depend on the model of the month, but on the ability to choose and change in a governed way.
The right setup starts with workload, data sensitivity and available hardware.
Open the hardware guide →