AI Model Comparison
Every major current model from OpenAI, Anthropic, and Google, side by side — pricing per million tokens and context window size. Sort or filter to find the right fit for a specific task.
Pricing and context windows last verified 2026-08-07 directly from each provider's own documentation — see sources below.
Context window, visualized
How much text each model can hold in a single request, relative to the largest context window in this table.
What context window actually means in practice
A context window is the maximum amount of text — measured in tokens, roughly 4 characters each in English — a model can consider in a single request, including both what you send it and what it generates back. A 1 million token window is roughly enough to hold a 1,500-page book, a sizeable codebase, or a very long chat history in one request. In practice, several independent reports have found that model accuracy can decline well before the stated maximum is reached, particularly past the first few hundred thousand tokens — worth keeping in mind for a task that depends on precise recall from deep in a long document, not just whether the content technically fits.
Sources
- OpenAI pricing: https://platform.openai.com/docs/pricing
- Anthropic pricing: https://docs.anthropic.com/en/docs/about-claude/pricing
- Google pricing: https://ai.google.dev/gemini-api/docs/pricing
Frequently asked questions
Why do context windows look so similar across providers now?
By 2026, roughly 1 million tokens has become a common flagship context window size across OpenAI, Anthropic, and Google's top models — a sharp jump from the 4K-32K windows common just a few years earlier. Actual usable accuracy near the top of that window can still vary between models, which a raw context window number doesn't capture.
Should I always pick the cheapest model?
Not necessarily — the cheapest model per token is only cheaper overall if it can actually complete the task correctly on the first try. A budget model that needs several retries, or produces output requiring heavy editing, can end up costing more in total time and tokens than a pricier model that gets it right the first time.
How often is this data updated?
The pricing and context window figures here are checked directly against each provider's own documentation periodically. The exact verification date is shown at the top of the table — for a purchasing decision, always confirm current pricing directly on the provider's site linked below.