# Claude Haiku 5.5 tops Artificial Analysis AA-Paperclip document benchmark at 92%

> Original title: AA-Paperclip tests models on extracting facts and making decisions about real-world documents. Each of its 1,500 private documents has fi...

- Source: Artificial Analysis
- Published: 2026-11-08T23:51:03.000Z
- Bad Influence: https://badinfluence.news/items/asxohyv43nzzuevuno6w5x8fo
- Original: https://x.com/ArtificialAnlys/status/2119578477259358404

## Summary

Artificial Analysis reports that Claude Haiku 5.5 (Max) passes 92% of documents on its AA-Paperclip benchmark, ahead of GPT-6 Luna (Max) at 86%. The benchmark requires a model to answer all 15 extraction and decision questions correctly on each of 1,500 private documents. Drex Max v1.1 leads the decision models at 63%, while Drex Max v1.1 answers 97% of individual questions correctly against Haiku's 99%.

## Post

AA-Paperclip tests models on extracting facts and making decisions about real-world documents. Each of its 1,500 private documents has five decisions and ten facts to extract. The key metric measures the share of documents where a model answers all 15 questions correctly. For the Decision Score, the model’s confidence in a document is the product of its 15 answer probabilities.

Claude Haiku 5.5 (Max) passes 92% of documents and GPT-6 Luna (Max) passes 86%. Drex Max v1.1 (63%) and GLiDE v1.2 (62%) lead the decision models, followed by Decider v1.2 (55%). The all-pass requirement separates models that look close on individual answers. Drex Max v1.1 gets 97% of answers right, close to Claude Haiku 5.5 (Max) at 99%, but passes 63% of documents against Haiku's 92%. Decision models lose most documents on the decisions rather than extraction, with Drex Max v1.1 correctly making all five decisions on 71% of documents and all ten extractions on 88%.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWpAbgSaMAAdZaI.jpg&mode=thumb&exp=1794528000&sig=abbd634d24989e08)