HELM AFR - Winogrande (Xhosa): leaderboard
The "Winogrande (Xhosa)" subtask of HELM AFR. Stanford HELM's African-languages project (mmlu-winogrande-afr): MMLU subject exams and Winogrande coreference translated into sixteen African languages, scored against the same items in English so a model's loss from the language switch...
Metric: EM. Source: crfm.stanford.edu. Saturation forecast: Estimated already saturated. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o (2024-08-06) | 67.8 | #335 |
| 2 | Claude 3.7 Sonnet (20250219) | 66.67 | #203 |
| 3 | Gemini 2.0 Flash | 64.23 | #344 |
| 4 | GPT-4 Turbo | 63.38 | #406 |
| 5 | Gemini 2.0 Flash Lite | 59.71 | #453 |
| 6 | DeepSeek V3 | 58.52 | #323 |
| 7 | Llama 3.1 405B Instruct | 54.05 | #461 |
| 8 | Llama 3.3 70B Instruct | 53.48 | #536 |
| 9 | Gemma 2 27B (IT) | 53.25 | #649 |
| 10 | GPT-3.5 Turbo (0125) | 52.86 | #988 |
| 11 | Mixtral 8x7B Instruct | 52.57 | #1172 |
| 12 | Claude 3.5 Haiku (20241022) | 52.46 | #591 |
| 13 | Qwen 2.5 72B Instruct | 52.12 | #458 |
| 14 | Llama 3.1 70B Instruct | 51.73 | #566 |
| 15 | Mistral 7B Instruct (v0.3) | 51.1 | #1276 |
No result here: #4 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=helm-afr-winogrande-xhosa · How It Works · Data refreshed daily, snapshot 2026-10-23.