HumanLikeness - Sound-2: leaderboard
HumanLikeness sound-association task comparing model responses with human sound-gender association patterns.
Metric: Humanlike Score (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 20 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | starchat2-15B-v0.1 | 70.12 | #832 |
| 2 | Phi-3 Mini 4K Instruct | 67.57 | #1297 |
| 3 | Llama 3.1 8B Instruct | 64.53 | #1035 |
| 4 | CodeLlama-34B-Instruct-hf | 63.97 | #1381 |
| 5 | GPT-4o | 62.86 | #343 |
| 6 | Mistral Nemo Instruct (2407) | 62.58 | #936 |
| 7 | Llama 3.1 70B Instruct | 61.97 | #561 |
| 8 | zephyr-7B-alpha | 61.83 | #1312 |
| 9 | Llama 3 70B Instruct | 61.36 | #634 |
| 10 | GPT-3.5 Turbo | 61.25 | #865 |
| 11 | Llama 2 7B Chat | 60.54 | #1499 |
| 12 | c4ai-command-r-plus | 59.93 | #724 |
| 13 | GPT-4o Mini | 58.34 | #599 |
| 14 | Mistral 7B Instruct (v0.3) | 58.32 | #1274 |
| 15 | Mistral 7B Instruct (v0.2) | 57.81 | #1333 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=humanlikeness-sound-2 · How It Works · Data refreshed daily, snapshot 2026-10-20.