List of AI Models Used Across EasySLR

List of AI Models Used Across EasySLR

EasySLR uses purpose-specific GPT models across different review stages to optimise accuracy, performance, and cost efficiency. Each model has been selected based on the complexity and nature of the task it supports, ensuring reliable AI assistance throughout the systematic review workflow.

Title–Abstract (TiAB) Screening
  • PICOS Extraction: GPT-5-mini
  • Screening Decisions:GPT-5-mini
This setup ensures fast and consistent interpretation of abstracts while maintaining strong alignment with PICOS criteria.

Full-Text (FT) Screening
  • PICOS Extraction: GPT-5.2
  • Screening Decisions: GPT-5.2
Data Extraction (DE)
  • Data Extraction & Reasoning: GPT-5.4
Search Query & Protocol Generation
  • Search Query Generation: GPT-5.2
  • Protocol Generation: GPT-5.2
Protocol Text Review
  • Protocol Review & Refinement: GPT-5.2
Ensures clarity, consistency, and methodological soundness in protocol text through advanced language understanding.

Conflict Analysis
  • Reviewer Conflict Analysis: GPT-5.2
Improves identification and interpretation of reviewer disagreements across screening and extraction stages.

Report Generation
  • Report Creation & Summarisation: GPT-5.2
Supports accurate, well-structured reporting for screening metrics, QC, and audit documentation.

PDF Upload Using AI
  • Smart PDF Matching: GPT-5.4-nano

Enables fast and efficient matching of PDFs.

Title & Abstract Mini Data Extraction AI

  • TIAB Mini Data Extraction: GPT-5-mini

 Supports rapid extraction of key study details from titles and abstracts to accelerate early-stage evidence synthesis workflows.

Full-Text Mini Data Extraction AI

  •  Full-Text Mini Data Extraction: GPT-5.2

 Enables structured extraction and summarisation of critical study information directly from full-text articles with improved contextual understanding.



Note: As of 15th January, these are the AI models currently in use. They are regularly updated, and any changes are also highlighted in the changelog.