> For the complete documentation index, see [llms.txt](https://scade.gitbook.io/scade-knowledge-base/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://scade.gitbook.io/scade-knowledge-base/flow-examples/5-minute-challenge-compare-different-llms.md).
# 5 minute challenge: compare different LLMs
Let's compare how different large language models are doing when encounter the same task
1. Go to and click the **New flow** button and then choose **Start with blank**
2. There will already be **Start** end **End** nodes on the canvas. In this particular case we don't need further integrations so we can simply delete the **End** node, but it's not a general rule.\
\
Click on the **End** node to highlight it and hit backspace
3. Go to the **Settings** of the **Start Node Form** and click **Configure fields** to start adding your data
4. Add one field, call it, for example, `prompt`. Click **Save**
5. Add your prompt to the **Start** node. For example, let's ask a question that would be easy for a human but can be tricky for AI: `how many N letters are there in some words, like "banana" or "sky"?`\
\
Hit the **Execute** button and then click **Save and execute** to run this node and generate the output. We need it to move further.
6. Use the search panel on the left to find processors that we will compare. Let's try **ChatGPT 3.5**, **ChatGPT 4**, **Claude 2** and **Claude 3**. \
\
Simply type name of the model (`ChatGPT` and `Claude`) in the panel and drag the processors to your workspace. We need two instances of the **ChatGPT** and two instances of the **Anthropic/Claude**
7. Once we have all four nodes of our LLMs, let's set models versions and name nodes accordingly\
\
Pick one of the **ChatGPT** nodes and go to the **Settings**. Locate the **Model** section and choose `gpt-4`
\
\
To rename the node hover over its title and click the pencil
\
\
Let's name this node **ChatGPT 4** for clarity
8. We don't have to change the model version of the next **ChatGPT** node, it has 3.5 set by default. Let's simply rename it to **ChatGPT 3.5**
9. Following suit, let's change the model version (for example, to `claude-3-opus`) of one of the **Anthropic/Claude** nodes and rename both of them to make it obvious
10. Connect the output of your **Start** node with the inputs of every other node
11. We are not done here yet. We need to go to each **ChatGPT** and **Claude** node and add messages. Go to the settings of the node, locate the **Messages** section, click the pencil and hit **Add message** after that\
\
\
If we work with LLM nodes, we need to use the **Expression editor** for obtaining data from other nodes. After adding the message, click the **#** symbol, and the **Expression editor** will open
\
\
Locate the **Start Node Form** on the left, click on it and drag the `prompt` to the expression field
\
\
Don't forget to click Save everywhere and repeat this course of action for all other **ChatGPT** and **Claude** nodes
12. All set! Hit the Play button on the top panel
13. Funny enough, **ChatGPT 3.5** done better than **ChatGPT 4** in this task. **Claude 3** is also completed the task
You can test any models in similar way and pick one that suits you best
