LLM model comparison

How to use it

1. Pick two to four models. Choose models from different providers or different tiers of the same one.
2. Describe your workload. Start from an example or enter your monthly requests and the tokens in and out of each one.
3. Compare and test. Read the table, then run your shortlist on real examples. Keep the cheapest model that clears your quality bar.

Comparing LLMs, explained

How do Claude, GPT, and Gemini compare on price?

Each provider sells a small and fast model, a mid-tier model, and a flagship, and prices rise steeply from one tier to the next. Prices are quoted per million tokens, with separate rates for input and output, so the cheapest model on paper isn’t always the cheapest for your workload. Pick the models above and enter your own volumes to see.

Why are input and output tokens priced differently?

Output costs more because the model generates it one token at a time, while input is processed in parallel. For tasks that read a lot and write a little, such as classification, the input price matters most. For drafting and code, the output price does.

What is a context window?

The most text a model can handle in one request, counting your input and its reply together. If a document plus the answer doesn’t fit, you have to split it up. The comparison shows whether one request fits in each model and how much of the window it uses.

What is cached input pricing?

When part of your prompt repeats across requests, such as a long system prompt or reference document, some providers charge a much lower rate to read it from cache. Set the cached share of your input and the comparison applies each model’s cached rate. Models without a modeled cache price are charged the full input rate.

What does batch pricing mean?

Some providers discount work that can wait, usually by 50% for jobs returned within 24 hours. It suits bulk jobs such as summarizing a backlog, not live chat. Check the batch box to apply it where the provider offers it.

Does the cheaper model give worse results?

Not necessarily, and the reverse can also be true. A small model is often enough for narrow, well-defined tasks, while hard reasoning favors a flagship. This tool compares price and capacity, not answer quality, so test your shortlist on real examples before you commit.

How current is the data?

Prices were checked against official provider pages on October 20, 2026, and context windows on October 21, 2026. Providers change prices and release models often, so confirm on the provider’s own pricing page before you sign a contract.

I’m Courtney Stephenson, an AI strategy and implementation leader, and I built this tool. Get in touch if you’re comparing models for a real project.