SGR-Bench - Goal-Oriented: leaderboard

Item-level F1 in percent for search agents on 100 state-gated retrieval tasks, where answer-bearing records on specialized data sites appear only after the right filters and selections are set (SGR-Bench, 2026), in the formulation that states only the goal and leaves the constraints implicit. Higher is better.

Metric: Item-level F1 (%) on the goal-oriented formulation of the same problems (only the goal is stated, constraints are implicit), 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 February 2027. 11 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GLM-5.168.1#198
2DeepSeek V4 Pro65.29#196
3GPT-5.564.15#54
4Gemini 3.1 Pro (Preview)Medium61.7#99
5Claude Opus 4.7Medium58.41#85
6Deep Research51.14
7Kimi K2.5Thinking45.17#266
8Gemini Deep Research30.15
9Qwen 3.6 Plus29.5#238
10Seed 2.0 ProMedium28.27#222
11Google Search AI Mode13.01

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=sgr-bench-goal-oriented · How It Works · Data refreshed daily, snapshot 2026-11-21.