Context window checker
Drop a file to see whether it fits the context window of the model you pick. You get a green or red answer, and a chunk preview when it is too long.
A context window is the most text an AI model can take in at once, counted in tokens. The context window checker converts your file to Markdown in your browser and estimates how many tokens it uses. It compares that estimate with the context window of the model chosen in the Model list. Green shows what percentage of the window the document fills. Red shows how many tokens it is over, with a link to a chunk preview you can export as JSONL. To compare models, choose another one from the list and the answer updates.
Frequently asked questions
Which context windows does the checker use?+
The published window for each model, in tokens: Claude Sonnet 5.5 1,000,000; Claude Opus 5.5 1,000,000; Claude Haiku 5.5 1,000,000; Claude Fable 5.1 1,000,000; GPT-6 Luna 1,050,000; GPT-6.1 Sol 1,050,000; GPT-6 Astra 1,050,000; Gemini 3.8 Flash 1,048,576; Gemini 3.1 Pro 1,048,576.
What happens if my document is too big?+
You get a red answer with the number of tokens over the limit and a link to the chunk preview. There you can set a chunk size below the model's window and export the chunks as JSONL.
How accurate is the fit answer?+
The token counts are estimates, marked approx, so a document very close to the limit may fit or not. OpenAI models are counted with o200k_base. For Claude the estimate assumes about 2.5 characters per token, and for Gemini about 4.