SGR-Bench - Row-F1: leaderboard
Metric: Row-level F1 (%) over all tasks: a row counts only when every field of the aligned row is correct, 100 state-gated retrieval tasks on specialized data sites (scholarly, official statistics, life science, environmental, regulatory); eight models in production CLI agents (GPT-5.5 in Codex CLI, the others in Claude Code) with the same Serper search, fetch and PDF tools, plus three commercial search agents; outputs parsed into rows and aligned by key fields; higher is better. Source: arxiv.org. Saturation forecast: Around September 2027. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 43.37 |
| 2 | Deep Research | 41.5 |
| 3 | Claude Opus 4.7 (Medium) | 38.51 |
| 4 | Kimi K2.5 (Thinking) | 33.74 |
| 5 | GLM-5.1 | 33.32 |
| 6 | DeepSeek V4 Pro | 32.64 |
| 7 | Gemini 3.1 Pro (Preview) (Medium) | 27.11 |
| 8 | Seed 2.0 Pro (Medium) | 20.45 |
| 9 | Qwen 3.6 Plus | 11.62 |
Interactive version: theaggregate.ai/benchmark?slug=sgr-bench-row-f1 · How It Works · Data refreshed daily, snapshot 2026-10-07.