Compare models

Model descriptions only tell you so much. The quickest way to find the best model for a piece of work is to give several models the same real prompt and read what comes back. Compare Models does exactly that: one prompt, two to five models, answers side by side, with the response time and credits for each.

app.stickyprompts.com/?isModelComparison=true Screen recording

Start a comparison

1. Open Compare Models

   Choose Compare Models from Apps in the sidebar, or from the What would you like to do? grid on the home screen.

   You can also switch on Comparison Mode at the bottom of the Select AI Model dialog. Open it from the model button in the composer, then Choose model.

2. Pick the models

   Select between two and five models. Mix makers and sizes: comparing a flagship with a fast model tells you whether the task really needs the flagship. See Choosing a model for how to read the model cards.

3. Write one prompt and send

   Write the prompt exactly as you would for real work, with the same context and attachments. Every selected model receives the same message.

   Write a 2-sentence product description for Northwind Analytics,
   a self-serve retail dashboard for chains with 10-200 stores.
   Plain English, no buzzwords.

Read the results

app.stickyprompts.com/conversations/D3CWS1GZ 1 Time and credits per model 2 Fastest and fewest credits 3 Export every answer 4 Back to one model

Each model gets its own column. At the top of each column you see the model’s name, how long it took to answer and the credits it used (1). The fastest model is highlighted.

Under the columns, a summary line (2) names the winner on each count, for example “Claude Opus 5.5 was fastest (0.7s) · GPT-6 Sol used fewest credits (25.04)”. Speed and credits are easy to measure; quality is your call, so read the answers themselves before you decide.

Each answer has the same action icons as a normal reply, so you can copy the one you like best. If a model used a tool along the way, such as a web search, you’ll see a note like 1 action above its answer, which you can expand.

Control	What it does
Download DOCX (3)	Downloads the comparison as a Word document, ready to share with colleagues who weren’t in the chat.
Statistics	Shows the numbers behind the comparison.
Exit comparison mode (4)	Returns the chat to a single model.

While comparison mode is on, the model button in the composer reads Comparing 3 models (or however many you picked), and the chat header shows the same.

When comparing pays off

• Choosing a default for a team. Before you standardise on a model for support replies or sales emails, compare the candidates on a handful of typical cases.
• High-stakes writing. For a board summary or a customer announcement, read three drafts and combine the best parts.
• New models. When a new model appears in the selector, compare it with the one you use today on your own prompts.
• Checking an answer. When a fact matters, asking several models and seeing whether they agree is a quick sanity check. For answers grounded in your own documents, use a knowledge base.

Each model in a comparison uses its own credits, just as it would in a separate chat. The per-model credits in each column show exactly what each answer used.

Related pages