Which AI model has the largest context window?

What does an 11-million-token context window hold?

Leiolai’s context includes the messages and supplied material in the conversation. With an experimental window of up to 11 million tokens, you can bring long document text or a large body of code into one task, then ask follow-up questions about the same material.

Which model can work with very long documents?

Leiolai 1 supports an experimental context window of up to 11 million tokens. That gives you room to bring substantial document text, code, or an extended conversation into a task.

Can I use the context window through the API?

Yes.

Send the material using the supported request format, including text and supported inline files.

Does a large context window also mean a long answer?

Context size and output length measure different things. Context is the material available to Leiolai; output is the response Leiolai writes. Experimental continuous generation provides no fixed output-token ceiling.

Which model ID provides Leiolai’s context window?

Use leiolai-1 in the developer API. The same model ID is available through Chat Completions, Responses, Anthropic Messages, and text Completions.

Is the 11-million-token context window experimental?

Yes.

Leiolai 1 supports an experimental context window of up to 11 million tokens.

How do I prepare a long document question?

Give Leiolai the document text and a question about that document. If the answer should rely only on the source, say so and ask for the supporting section or passage. For a comparison, name the documents and the criteria to compare.

Can developers supply text files as context?

Leiolai Chat Completions and Responses accept supported inline UTF-8 text or code files, up to 2 MiB per file. Send the supported content representation rather than a binary PDF or a file ID. Include the task in the accompanying text.

Does conversation history count toward the input?

Earlier messages sent with a Leiolai API request form part of the context. Stateless conversation formats need the relevant history sent again on the next turn. That history also contributes to input token usage.