GPT-5.5 reference generation improves with higher reasoning effort

A new study published on arXiv investigated the impact of increased reasoning effort on the reference generation capabilities of GPT-5.5. Researchers found that when GPT-5.5 operated at a higher reasoning effort, it improved the verifiability of generated scholarly references from 78.3% to 84.8%. The study also observed that higher reasoning effort resulted in longer reasoning summaries from the model, with predefined features scoring higher, though discovered features showed less uniform changes. AI

IMPACT This research suggests that higher computational effort in LLMs can lead to more reliable outputs in academic contexts, potentially improving the efficiency of scientific research.

RANK_REASON Research paper detailing model capabilities and findings. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• Bibliographic Explorer
• CatalyzeX Code Finder for Papers
• Connected Papers
• CORE Recommender
• DagsHub
• Gotit.pub
• GPT-5.5
• Hugging Face
• Influence Flower
• Litmaps
• ScienceCast
• scite Smart Citations

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →