Skip to main content
No single AI model is best at everything. Claude excels at nuanced reasoning and long-form synthesis; GPT 5.5 shines at structured problem-solving and code; Gemini 3.1 Pro brings strong multimodal and analytical capabilities; Grok offers a different perspective shaped by real-time information. Noesis gives you access to all of them in one workspace — and lets you switch between models mid-conversation, so you can direct each message to whichever model is best suited for that moment without starting over in a new chat.

​
Available Models by Plan

Origin (Free)

Standard models included at no cost:
  • Gemini 3.5 Flash
  • Claude Haiku 4.5
  • GPT 5.4 Mini
  • Grok 4.3
These models are fast, capable, and well-suited for research drafts, Q&A, summarization, and general exploration.

Synthesis & Noesis

Premium models for deep analysis:
  • Claude Opus 4.8
  • GPT 5.5
  • Gemini 3.1 Pro
  • And more as they are released
Premium models offer significantly deeper reasoning, larger context windows, and stronger performance on complex, multi-step research tasks.

​
Switching Models Mid-Conversation

To change the active AI model, click the model selector dropdown at the bottom of the chat input field. A menu appears listing all models available on your plan. Select any model and your next message will be sent to it. Switching models does not interrupt your conversation. The new model receives the full conversation history up to that point and picks up exactly where the previous model left off — no re-introduction or context re-establishment needed. This makes it seamless to, for example, use a fast model to sketch out an idea and then switch to a more powerful model to deepen the analysis.

​
Comparing Models on the Same Branch

The most powerful way to evaluate different models is to pit them against each other on the same question. Here’s the recommended pattern:
  1. Ask your question in the main thread using your baseline model.
  2. Branch the conversation from that message.
  3. Switch to a different model on the new branch.
  4. Ask the same question (or the AI’s follow-up) and observe the difference in reasoning, tone, and conclusions.
  5. Synthesize both branches to produce a unified, model-agnostic summary.
This gives you genuine signal about how different architectures approach the same problem, rather than anchoring on one model’s output.

​
Custom System Prompts Per Branch

You can define a custom system prompt for each branch independently. This means you can, for example, instruct Claude Opus 4.8 to respond as a domain expert on one branch, and GPT 5.5 to respond as a skeptical peer reviewer on another — then synthesize both outputs. System prompts are set in the branch settings panel and override the default workspace prompt for that branch only. See the Branching documentation for full details on per-branch configuration.
Use faster, lighter models (Gemini 3.5 Flash, Claude Haiku 4.5, GPT 5.4 Mini) for early-stage drafting, brainstorming, and exploratory questions where speed matters more than depth. Once you’ve identified the key research direction, switch to a premium model (Claude Opus 4.8, GPT 5.5, Gemini 3.1 Pro) to do the heavy analytical lifting. This keeps your workflow fast without sacrificing quality where it counts.
Premium models (Claude Opus 4.8, GPT 5.5, Gemini 3.1 Pro) consume your plan’s usage allowance at a significantly higher rate than standard models. If you are on the Synthesis or Noesis plan, monitor your usage indicator in the sidebar to avoid hitting your limit mid-session. The Noesis plan includes 5× the usage of Synthesis, making it the right choice for heavy premium model workflows.