Compare AI Responses sends the same prompt to several AI models and shows their outputs side by side. It has two modes: Text Responses, where models answer a question, and Image Generation, where models each generate an image from the same description.
How a comparison works
- Choose the mode. Text or image.
- Pick the models. The model list is grouped by provider: OpenAI GPT, Google Gemini, Anthropic Claude, DeepSeek, and xAI Grok models in text mode; OpenAI and Google image models in image mode. The exact versions on offer change as providers release them.
- Write one prompt. A question for text mode, or a description of the image you want.
- Send. Every selected model receives the same prompt at once, and the answers appear in columns.
Credits
Credit use varies by model. The page shows a Credit usage by model table with the maximum credits per request for each model, and above the prompt it shows the expected cost for your current selection: "Up to N credits, charged by answer length." Text answers are charged by their length up to that maximum; image generations are charged per image. A simple way to budget: one credit per model is the floor, and the table tells you the ceiling.
Two ways educators use this tool
For course preparation: Before using an AI explanation in a lesson, compare how different models explain the same concept. One model may produce a cleaner analogy; another may be more technically precise. Seeing the options helps you choose the best explanation, or write a better prompt by seeing where each model falls short. Image mode does the same for a diagram or illustration you want generated for a slide.
For AI literacy teaching: Comparison activities are one of the most effective ways to teach learners to think critically about AI output. When learners see two different answers to the same question, they have to evaluate rather than accept. Inside a course, the Chat with AI lesson type gives learners the same comparison directly.
What to evaluate when comparing
| What to compare | Why it matters |
|---|---|
| Accuracy | Does the answer hold up? Would a subject expert correct it? |
| Explanation quality | Which answer would a learner actually understand? |
| Reasoning transparency | Does the model show its work, or just give a conclusion? |
| Hallucination risk | Which answer makes confident claims that are hard to verify? |
| Tone and audience fit | Which response matches the level and context you teach in? |
| In image mode: fidelity | Which image shows what the prompt asked for, with the labels and proportions a lesson needs? |
Avoid comparing only for fluency. A polished, well-structured response is not necessarily a correct or useful one. Models can agree with each other and still be wrong. The question is always: which answer best serves the learner?
Building comparison lessons
If you are designing an AI literacy course or a module on critical AI evaluation, the tool gives you real, current output to work with rather than curated examples.
A strong comparison lesson follows this structure:
- Give learners a prompt to test themselves.
- Show the comparison output, or have learners generate it in a Chat with AI lesson or in TutorCampus.
- Ask learners to rank the responses and explain their reasoning.
- Reveal common issues in each response: hallucinations, missing nuance, style mismatches.
- Have learners revise the prompt and observe what changes.