BETA
Index · Side-by-Side Comparison
Pick up to 4 models to compare on pricing, context, capabilities and benchmarks. A free AI model comparison tool for LLM pricing, context windows and benchmark scores side by side. Search by model name or provider; typos are forgiven. See full per-token rates on the AI pricing index, or use the cost calculator to estimate spend for a specific prompt or volume workload.
A fair model comparison lines up the same metrics across every candidate. The ones that move a decision are input price and output price (per million tokens, billed separately because output is usually costlier), the blended cost that weights those two rates into one number for quick ranking, the context window that caps how much text a model can read and reason over in a single request, and the benchmark scores that proxy capability on reasoning, coding, and knowledge tasks. To read the table, scan one row at a time: each row is a single attribute and each column is a model, so the leader on any metric is easy to spot. The cheapest model is rarely the cheapest on every row — a low input rate can be offset by an expensive output rate or a smaller context window — so weight the rows that match your workload. Use price and context to set a shortlist, then let benchmarks break the tie. How the blended cost is calculated →
Add each model to the comparison, then read them column by column across the same metrics: input and output price per million tokens, blended cost, context window, modalities, and benchmark scores. Comparing on identical rows keeps the choice objective, pick the model that wins on the metrics that matter for your workload rather than on headline reputation.
Blended cost combines a model's separate input and output token rates into a single per-million-token figure using an assumed mix of input to output tokens (commonly 70% input, 30% output). It makes models quick to rank side by side, but because real workloads vary, you should still check the raw input and output prices for your own token ratio.
The cheaper model depends on your token mix. Compare the input price, output price, and blended cost rows: a model with a lower input rate may still cost more overall if your workload is output-heavy, since output tokens are usually priced higher. For a precise answer, run your expected input and output volumes through the cost calculator.