Tool

Context window visualizer.

"1M context" sounds abstract until you see it as 1,500 pages or eight novels. Pick models and see what their windows actually hold.

Rules of thumb, with the math shown: one page ≈ 670 tokens; an average novel ≈ 120,000 tokens; a 400-line code file ≈ 4,000 tokens; one hour of meeting transcript ≈ 9,000 tokens. Bars use a square-root scale so 8K and 2M windows can share one chart.