source
Scale Labs
Benchmarks · /benchmarks/scale-tutorbench
TutorBench
Scale tutoring evaluation on math, science, and instructional guidance quality.
Passport
Visible tradeoffsThis is a rubric-judged signal, so it is more structured than arena taste but still depends on the scoring rubric.
metric
Score (%)
judge
Rubric
direction
higher better
group id
scale_tutorbench_current
domain
Reasoning / math / science
What it measures vs what it misses
✓ Measures
How well a model tutors through multi-step academic problems. Instruction quality, pedagogy, and reasoning support on teaching-style prompts.
✗ Misses
Live classroom preference. Latency and cost.
Why this countsIt is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.Comparable-group ruleThis percentile only compares models inside the exact benchmark/version group shown here. It is not a universal score.What it missesIt still misses product usability, latency, and whether the model stays correct in messy real workflows.
Leaderboard · this benchmark version
#1 · GPT-5.4
SL · Apr 29, 2026
56.6%
#2 · Gemini 2.5 Pro
SL · Apr 29, 2026
55.7%
#3 · GPT-5
SL · Apr 29, 2026
55.3%
#4 · GPT-5.4 mini
SL · Apr 29, 2026
55.3%
#5 · GPT-5.4 nano
SL · Apr 29, 2026
55.3%
#6 · GPT-5.1
SL · Apr 29, 2026
54.1%
#7 · Gemini 3 Pro Preview
SL · Apr 29, 2026
53.7%
#8 · Claude Opus 4.6
SL · Apr 29, 2026
53.6%
#9 · Gemini 3.1 Pro
SL · Apr 29, 2026
53%
#10 · Gemini 3.1 Flash-Lite Preview
SL · Apr 29, 2026
51.5%
#11 · Llama 4 Maverick
SL · Apr 29, 2026
40.2%