HELM AFR - Winogrande (Tswana): leaderboard
The "Winogrande (Tswana)" 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 | Gemini 2.0 Flash | 71.99 | #345 |
| 2 | Claude 3.7 Sonnet (20250219) | 71.59 | #205 |
| 3 | GPT-4o (2024-08-06) | 67.74 | #337 |
| 4 | Gemini 2.0 Flash Lite | 64.57 | #455 |
| 5 | GPT-4 Turbo | 64.12 | #410 |
| 6 | DeepSeek V3 | 59.76 | #325 |
| 7 | Llama 3.1 405B Instruct | 55.86 | #464 |
| 8 | Gemma 2 27B (IT) | 54.27 | #653 |
| 9 | Claude 3.5 Haiku (20241022) | 52.86 | #591 |
| 10 | Llama 3.1 70B Instruct | 52.8 | #567 |
| 11 | Llama 3.3 70B Instruct | 52.69 | #537 |
| 12 | GPT-3.5 Turbo (0125) | 52.24 | #998 |
| 13 | Qwen 2.5 72B Instruct | 52.01 | #463 |
| 14 | Mixtral 8x7B Instruct | 51.16 | #1171 |
| 15 | Mistral 7B Instruct (v0.3) | 50.76 | #1272 |
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-tswana · How It Works · Data refreshed daily, snapshot 2026-10-24.