You can let Coworker assign the best model, or manually select a specific one if you have a preference or want to experiment.
How to choose your model
Go to the Work page.
Find the model dropdown in the chat input bar.
Click it to see all available models.
Select Optimized to let Coworker select the most suitable model that balances quality and cost, or choose a specific model.
Note: Your model selection applies to the current session. Starting a new chat resets the selector to Optimized (the default).
What is "Optimized"?
Optimized is the default setting and the recommended choice for most users. When selected, Coworker routes your task to the most appropriate model from its portfolio — balancing output quality and cost efficiency behind the scenes.
This means your tasks get handled by fast, cost-efficient models (such as GLM 5.3 Flash), whilst more complex reasoning or writing tasks can be manually assigned to frontier models like Claude Sonnet or Claude Opus. You get frontier-quality results without paying for heavy compute on every message.
Note: All model routing under Optimize uses US-hosted infrastructure.
Available models
Coworker offers a mix of frontier AI models such as Claude, ChatGPT, Gemini, Kimi and GLM. New versions of models will be automatically made available within Coworker as they are released.
All models work with your organizational data via Coworker's OM2 layer — so the core context-aware experience is consistent regardless of which model you select.
Which model should I use?
When in doubt, leave it on Optimized. If you want to manually choose:
Task type | Recommended model |
General research, Q&A, and analysis | Optimize or Claude Sonnet |
Complex reasoning or nuanced writing | Optimize or Claude Opus |
Structured outputs and data formatting | Optimize or Claude Sonnet |
Processing large amounts of text | Optimize or Gemini |
High-volume or cost-sensitive tasks | Optimize or GLM 5.3 Flash |
You have a model preference from outside Coworker | Select whichever you know best |
Tips
Optimized is the default and recommended setting. It delivers the best balance of quality and cost without requiring any configuration.
All models work in both Chat and Build modes. You can switch models regardless of which mode you're using.
Your selection persists within a session. If you switch to ChatGPT mid-conversation, Coworker will continue using it for the rest of that chat until you change it again or start a new chat.
If a model feels slow or unresponsive, try switching to another. You can also return to Optimized, which includes built-in fallback routing.