TutorBench: leaderboard
TutorBench evaluates how well LLMs perform common tutoring tasks for high school and AP-level subjects.
Metric: Score. Source: labs.scale.com. Source: scale.com. Source: benchmarklist.com. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Muse Spark | 68.55 | #64 |
| 2 | Gemini 2.5 Pro (Preview 06-05) | 55.65 | #106 |
| 3 | GPT-5 | 55.33 | #98 |
| 4 | Kimi K2.5 | 54.56 | #144 |
| 5 | GPT-5.1 (Thinking) | 54.09 | #135 (GPT-5.1) |
| 6 | Claude Opus 4.6 (Adaptive Reasoning, Max Effort) | 53.68 | #63 (Claude Opus 4.6) |
| 7 | Gemini 3 Pro (Preview) | 53.67 | #66 |
| 8 | Claude Opus 4.6 (Non-reasoning) | 53.55 | #63 (Claude Opus 4.6) |
| 9 | GPT-5.2 | 53.49 | #110 |
| 10 | Gemini 3.1 Pro (Preview) | 52.99 | #58 |
| 11 | O3 (2025-04-16) (Medium) | 52.76 | #121 (O3 (2025-04-16)) |
| 12 | O3 (2025-04-16) (High) | 52.09 | #121 (O3 (2025-04-16)) |
| 13 | Gemini 3.1 Flash Lite (Preview) | 51.5 | #194 |
| 14 | Claude Opus 4.5 (20251101) (Thinking) | 51.2 | #74 (Claude Opus 4.5 (20251101)) |
| 15 | Claude Opus 4.1 (20250805) (Thinking) | 50.78 | #132 (Claude Opus 4.1 (20250805)) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=tutorbench · How It Works · Data refreshed daily, snapshot 2026-10-20.