HELM AFR - Overall (Xhosa): leaderboard

Metric: Mean score. Source: crfm.stanford.edu. 22 models tracked.

Top models

#ModelScore
1GPT-4o (2024-08-06)68
2Gemini 2.0 Flash67.94
3Claude 3.7 Sonnet (20250219)66.72
4Gemini 2.0 Flash Lite63.93
5DeepSeek V360.86
6GPT-4 Turbo58.37
7Llama 3.1 405B Instruct49.8
8Gemma 2 27B (IT)48.52
9Llama 3.3 70B Instruct47.74
10Llama 3.1 70B Instruct47.65
11Claude 3.5 Haiku (20241022)45.74
12Qwen 2.5 72B Instruct43.59
13Gemma 2 9B (IT)42.55
14GPT-3.5 Turbo (0125)39.83
15Llama 3.1 8B Instruct37.19

Interactive version: theaggregate.ai/benchmark?slug=helm-afr-overall-xhosa · How It Works · Data refreshed daily, snapshot 2026-09-08.