Reasoning & Verbosity
Suprvisr’s upgraded language model supports multi-level reasoning — it can think more deeply, source more documentation, and build responses you can act on with greater confidence. Alongside this, verbosity modes let you decide how much depth you want: a quick summary or a thorough, detailed response.
Multi-level reasoning
When you ask a complex question, Suprvisr doesn’t just jump to an answer. With multi-level reasoning enabled, the AI:
- Breaks down the problem — identifies what it needs to answer fully
- Sources documentation — pulls from your connected data sources and internal knowledge
- Builds step-by-step — constructs the answer in layers, checking each step
- Surfaces its work — you can see the reasoning process as it happens (in the Thinking panel)
This is especially useful for:
- Complex analytical questions
- Multi-step technical problems
- Answers that require reconciling multiple documents
- High-stakes decisions where you need to understand the reasoning, not just the conclusion
For simple questions, reasoning adds overhead. Save deeper reasoning for questions where accuracy and thoroughness really matter.
Verbosity modes
Verbosity controls how long and detailed the AI’s response is.
| Mode | Best for | What you get |
|---|---|---|
| Quick | Fast answers, simple questions | Short, direct, no fluff |
| Balanced | Most everyday use | Clear answer with enough context |
| Verbose | Deep research, complex topics | Long, thorough, fully explained |
Quick mode
Gets out of the way. No lengthy preambles, no restating the question, no caveats — just the answer. Ideal when you know what you’re asking and need a fast response.
Balanced mode
The default. Provides a direct answer with enough surrounding context to be useful and actionable.
Verbose mode
For when you need the full picture. Long-form, well-structured responses that explain reasoning, cover edge cases, and cite sources. Use this for research, reports, or any output you’ll share with others.
Reasoning visibility
Regardless of verbosity mode, you can always see the AI’s reasoning in the Thinking panel during generation. This shows:
- What the AI is working through
- What sources it’s pulling from
- Observation results from tool calls
Reasoning visibility helps you catch errors early — if the AI is reasoning from a wrong assumption, you can stop it and redirect before it finishes.
Related
- AI Chat — the main chat interface
- Citations — see sources used in deep-reasoning responses
- Prompt Strategies — how to ask questions that get the best answers