01 / CONTINUITY
01 / CONTINUITY
Keep useful work close.
Reuse working data while the chat’s Python runtime is alive. Keep saved files and approved memories separately, with clear limits on what survives a restart.
ABOUT / WHY VARIANT-1
Make room
for the work.
AI work often continues long after the first answer. VARIANT-1 is being built for that continuation: a place where models can use code, keep working data, and adapt their tools as a task develops.
01 / PURPOSE
Good work
takes more than a turn.
You compare a set of files, spot an exception, and ask a different question. You build a report, open it, and notice something that needs changing. Each step depends on work already done.
VARIANT-1 brings those steps into one workspace. Its live Python environment keeps data and methods available within the chat, while tools connect the task to files, the web, and desktop applications.
02 / ADAPTATION
02 / ADAPTATION
Make the method visible.
Let the model combine tools and revise its approach. Read the code, inspect the output, and use the result to guide the next step.
03 / CHOICE
03 / CHOICE
Choose for the task.
Choose a supported local model or cloud connection. Keep one workspace while weighing each model’s capabilities, cost, and requirements.
LIVING MATTER / STUDY 01
Keep what helps.
Change what needs to change.A DESIGN PRINCIPLE / VARIANT-1
02 / APPROACH
Build around
what helps the work.
Models differ in how they plan, use tools, and recover from mistakes. VARIANT-1’s approach is to give them a common working environment, with guidance that can be revised as their capabilities change.
The aim is practical: spend less effort reconstructing the task and more effort moving it forward. That means testing real workflows, examining failures, and improving the parts that get in the way.
Progress has to show up in the result: a file that opens, a calculation that checks out, or a change you can see in the application.
03 / CURRENT DIRECTION
A Windows workspace.
Your choice of model.
The public beta is being prepared for people who want to put AI to work across code, research, files, and everyday desktop tasks. The source is available as an early public preview under the MIT License. A free Windows installer is still in preparation.
Bring a supported local model, API key, or provider account. The beta requires no VARIANT account or subscription. Your provider’s prices and access requirements still apply.
Read the release status