HELM AFR - Overall (Tsonga): leaderboard
Metric: Mean score. Source: crfm.stanford.edu. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.0 Flash | 63.15 |
| 2 | Claude 3.7 Sonnet (20250219) | 60.75 |
| 3 | GPT-4o (2024-08-06) | 58.67 |
| 4 | Gemini 2.0 Flash Lite | 56.86 |
| 5 | DeepSeek V3 | 55.22 |
| 6 | GPT-4 Turbo | 51.69 |
| 7 | Llama 3.1 405B Instruct | 49.14 |
| 8 | Llama 3.3 70B Instruct | 47.19 |
| 9 | Llama 3.1 70B Instruct | 46.41 |
| 10 | Claude 3.5 Haiku (20241022) | 42.1 |
| 11 | Gemma 2 27B (IT) | 41.79 |
| 12 | Qwen 2.5 72B Instruct | 41.44 |
| 13 | GPT-3.5 Turbo (0125) | 39.11 |
| 14 | Llama 3.1 8B Instruct | 37.32 |
| 15 | Mistral 7B Instruct (v0.3) | 36.21 |
Interactive version: theaggregate.ai/benchmark?slug=helm-afr-overall-tsonga · How It Works · Data refreshed daily, snapshot 2026-09-08.