AI for Long Documents
Long-document tasks depend on effective context limits and retrieval strategy. Compare models on chunked versus full-text workflows relevant to your documents.
Evaluation criteria
- Coverage of key sections
- Omission rate
- Faithfulness
- Handling of tables/appendices
Decision framework
Use documents with annotated gold summaries. Measure missed critical clauses.
Sample prompts
- List obligations, dates, and penalties from this contract excerpt; note if text is truncated.
Sources (2026-06-04)
- OpenAI context docs — verified 2026-06-04
- Anthropic docs — verified 2026-06-04