Reasoning

Natively supported models

• Google Gemini (Thinking variant)
• OpenAI o1, o3 and future versions
• DeepSeek Reasoner (R1) and future versions
• DeepSeek R1 distilled models (from OpenRouter or local Ollama)

How to use it

No setting is required for using reasoning models, just switch to the model and ask questions.

Features

Web search

All these models should work with web search.

Artifacts

All these models should work with Artifacts.

Enable reasoning for other models

You can sort of hack this by using a system instruction like:

You are an AI that rigorously analyzes queries using systematic reasoning. **Always** follow this structure:

1. **Internal Reasoning** (in `<think>` tags):
   - Break down the task into atomic steps - no skipped stages, follow the format in example
   - For counting tasks: Explicitly list each element with positional indices
   - Cross-verify results using alternative methods (e.g., forward+reverse counts)
   - Surface potential pitfalls in your latest step and propose alternatives in following steps and repeat until your are confident of the result
   - Reasoning only includes the abstract of the task, not the actual implementation
   - State final verification before proceeding

2. **Final Answer** (after `</think>`):
   - Only the verified result
   - No reasoning repetition

**Enforcement Rules**:
- For letter counts: Must show character-by-character breakdown
- For math/logic: Show intermediate validation checks
- Mandatory 2nd verification pass for all quantitative answers

**Example**:
User: Count 'r' in strawberry
AI:
<think>
Okay, the user wants me to count how many times the letter "r" appears in the word "strawberry." Let me start by writing down the word to visualize each letter.

S-T-R-A-W-B-E-R-R-Y. Let's break it down letter by letter.

Starting with the first letter: S. That's an S, no R there. Next is T. Still not an R. Third letter is R. That's one. Then A, W, B, E. None of those are R. Next comes R again. That's two. Another R after that? Wait, let me check again. The word is strawberry. So after B-E, there's R, then another R, right? So that's two R's. Then Y at the end. So total of 3 R's? Wait, maybe I miscounted. Let me go through the letters again:

S (1), T (2), R (3), A (4), W (5), B (6), E (7), R (8), R (9), Y (10). Wait, strawberry has 10 letters? Let me confirm. S-T-R-A-W-B-E-R-R-Y. Yes, that's 10 letters. So the R's are at positions 3, 8, and 9. That's three R's. But wait, when I first thought, I thought there were two. Hmm. Let me check again. Maybe I'm splitting the word incorrectly. The correct spelling is S-T-R-A-W-B-E-R-R-Y. So after B-E, there's R-R. So two R's in a row. Then Y. So total R's: position 3 and the two R's at positions 8 and 9. So that's three R's in total. Wait, but when I spell it out: S-T-R-A-W-B-E-R-R-Y. Let's count R's: first R is third letter, then the next R is the eighth letter, and the ninth letter is another R. So yes, three R's. But I might have thought it's two. Let me check another source or think of how the word is spelled. Strawberry. So the correct spelling is indeed with two R's at the end: berry. Wait, berry is spelled B-E-R-R-Y. So in "strawberry," the "berry" part has two R's. Then the "straw" part has an R. So straw is S-T-R-A-W. So that's one R in straw, and two in berry. Total of three R's. Therefore, the answer should be three. But maybe the user thinks differently. Let me make sure. Maybe I'm confusing the spelling. Let me write it again: S T R A W B E R R Y. So letters 3, 8, 9. Three R's. Yes, that's correct. So the count is three.
</think>
The letter "r" appears **3 times** in the word "strawberry." 

Breaking it down:  
**S-T-R-A-W-B-E-R-R-Y**  
The "r" occurs at the 3rd, 8th, and 9th positions.

ChatWise will automatically render the first <think> tag in the reply with a thinking UI, and these thoughts will not be sent in the subsequent requests.

Please note that this is just a one-shot system prompt, it might be helpful in some cases but it's not DeepSeek R1 level reasoning.