StudentBench - Learning Gain: leaderboard

Metric: GRE learning gain after one hour of AI tutoring (percentage points, pretest-adjusted ANCOVA). Source: studentbench.org. Saturation forecast: Around March 2028. 12 models tracked.

Top models

#ModelScore
1Claude Opus 4.8 (xHigh)16
2Gemini 3.1 Pro (Preview) (High)15.4
3Gemini 3.5 Flash (Low)14.3
4Gemini 3.7 Flash (Medium)14.1
5Claude Opus 4.8 (Non-reasoning)13.6
6GPT-5.5 (High)13.5
7Gemma 4 31B (High)13.1
8GPT-5.4 Mini (Non-reasoning)13
9Claude Opus 5 (High)12.6

Interactive version: theaggregate.ai/benchmark?slug=studentbench-learning-gain · How It Works · Data refreshed daily, snapshot 2026-10-04.