HumanLikeness - Word-2 — leaderboard

Metric: Humanlike Score (%). Source: huggingface.co. 20 models tracked.

Top models

#ModelScore
1zephyr-7B-beta25.64
2Llama 2 13B Chat Base25.05
3Mistral 7B Instruct (v0.3)25
4starchat2-15B-v0.124.52
5CodeLlama-34B-Instruct-hf23.32
6zephyr-7B-alpha22.88
7Mixtral 8x7B Instruct (v0.1)22.7
8Llama 2 7B Chat22.13
9Phi-3 Mini 4K Instruct19.26
10Llama 3 8B Instruct14.47
11Mistral 7B Instruct (v0.2)13.73
12Yi 1.5 34B Chat12.63
13Llama 3.1 8B Instruct11.81
14Mistral Nemo Instruct (2407)9.86
15c4ai-command-r-plus8.24

Interactive version: theaggregate.ai/benchmark?slug=humanlikeness-word-2 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.