PitchBench - Chord Pitch Count: leaderboard
Pitch hearing in audio-language models (PitchBench, Liessens Dujardin et al., 2026): counting the distinct pitches sounding together in a chord or random pitch set (228 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 (%; count the simultaneous pitches of a chord). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 46.5 | #91 |
| 2 | Gemini 3 Flash | 39.9 | #159 |
| 3 | Qwen3.5 Omni Flash | 20.2 | #548 |
| 4 | Qwen 3.5 Omni Plus | 9.7 | #237 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=pitchbench-chord-pitch-count · How It Works · Data refreshed daily, snapshot 2026-11-12.