BlueFin - Synthesis: leaderboard

Metric: Criteria-weighted rubric score (%; points met over maximum points of expert rubrics checked by an agentic GPT-5.4 judge for synthesis and manipulation workbooks, answer correctness for interrogation, non-completions count as zero; 20-tool spreadsheet harness, high reasoning effort; 9 held-out tasks building a financial model from a brief). Source: arxiv.org. Saturation forecast: Around September 2027. 5 models tracked.

Top models

#ModelScore
1Claude Opus 4.7 (High)66.7
2GPT-5.5 (High)59.9
3Claude Sonnet 4.6 (High)53.2
4Gemini 3.1 Pro (Preview) (High)50.3

Interactive version: theaggregate.ai/benchmark?slug=bluefin-synthesis · How It Works · Data refreshed daily, snapshot 2026-09-26.