The detail
A 2,000-word document is the length of a substantial blog post, a detailed product spec, or a short report. At about 2,660 tokens, it uses less than 2% of even a 200,000-token context window, so the practical question is cost and response quality, not fit.
Cost scales linearly: at $1 per million input tokens the document costs about $0.0027 to send once, but the same document included in every message of a 20-turn conversation is re-read each turn — roughly 53,000 cumulative input tokens. That is why long references belong in cached prompts or retrieval rather than pasted into every message.
For summarization tasks, expect the output to add its own tokens: a one-page summary of a 2,000-word document adds roughly 500-700 output tokens at the higher output rate.