PitchBench - Pitch Ranking: leaderboard
Pitch hearing in audio-language models (PitchBench, Liessens Dujardin et al., 2026): ranking three to seven tones from lowest to highest (120 stimuli), with synthetic stimuli. A pitch answer counts when any of MIDI number, scientific pitch notation or frequency is right; accuracy in percent, higher is better.
Metric: Accuracy (%; rank three to seven tones from lowest to highest). Source: arxiv.org. Saturation forecast: Around August 2028. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 Omni Plus | 20 | #237 |
| 2 | Gemini 3.1 Pro (Preview) | 5 | #91 |
| 3 | Qwen3.5 Omni Flash | 2.5 | #548 |
| 4 | Gemini 3 Flash | 0 | #159 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=pitchbench-pitch-ranking · How It Works · Data refreshed daily, snapshot 2026-11-12.