Skip to main content

​
Available Models

Cerebrum integrates with leading AI providers, giving you access to the most powerful models available:
ModelProviderAvailable From
Gemini 3 ProGoogleAll tiers
GPT-5.1OpenAIStarter+
Claude Sonnet 4.5AnthropicStarter+
Claude Opus 4.5AnthropicPro only
Grok 4.1xAIPro only

​
How to Switch Models

Model switching limitations: If your chat started with Gemini, you cannot switch to other models. However, if you started with any other model (GPT, Claude, Grok), you can freely switch between all models, including Gemini.
  1. Open the Copilot panel in the Canvas
  2. Click on the model name at the bottom of the chat
  3. Select your preferred model from the dropdown
  4. Continue your conversation with the new model
Your conversation history is preserved when switching models.

​
Choosing the Right Model

Best for: General tasks, multimodal inputs, large context windowsGoogle’s latest model with excellent performance across a wide range of tasks. Great starting point for most projects.
Best for: Complex reasoning, advanced code generation, nuanced instructionsOpenAI’s flagship model with superior reasoning capabilities. Ideal for complex agentic applications.
Best for: Long documents, detailed analysis, nuanced writingAnthropic’s balanced model offering great performance at a reasonable cost. Excellent for code review and documentation.
Best for: Most complex tasks, research-level reasoning, creative solutionsAnthropic’s most capable model. Use when you need maximum intelligence for challenging problems.
Best for: Real-time information, unconventional approaches, rapid iterationxAI’s latest model with unique capabilities and perspectives. Great for brainstorming and exploration.

​
Model Costs

Different models have different costs per request. Higher-tier models typically cost more but may complete tasks in fewer iterations.
Gemini 3 Pro is a great starting point and works well for UI-focused tasks. However, if you find yourself stuck, try starting a new chat with a different model - you might get better results for complex reasoning or code generation.