HELM AFR - Winogrande (Zulu): leaderboard
The "Winogrande (Zulu)" 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) | 70.18 | #335 |
| 2 | Gemini 2.0 Flash | 68.14 | #344 |
| 3 | GPT-4 Turbo | 65.76 | #406 |
| 4 | Claude 3.7 Sonnet (20250219) | 63.84 | #203 |
| 5 | Gemini 2.0 Flash Lite | 63.38 | #453 |
| 6 | DeepSeek V3 | 60.44 | #323 |
| 7 | Llama 3.1 405B Instruct | 56.14 | #461 |
| 8 | Gemma 2 27B (IT) | 54.1 | #649 |
| 9 | Qwen 2.5 72B Instruct | 52.29 | #458 |
| 10 | Llama 3.1 70B Instruct | 52.07 | #566 |
| 11 | Llama 3.3 70B Instruct | 51.95 | #536 |
| 12 | GPT-3.5 Turbo (0125) | 51.05 | #988 |
| 13 | GPT-2 XL | 50.65 | #1617 |
| 14 | Mixtral 8x7B Instruct | 50.59 | #1172 |
| 15 | Mistral 7B Instruct (v0.3) | 50.31 | #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-zulu · How It Works · Data refreshed daily, snapshot 2026-10-23.