SAHARA - POS Tagging: leaderboard
Part-of-speech tagging on texts in African languages, from the SAHARA African-language LLM leaderboard (UBC-NLP, ACL 2025). The task score is the mean over the languages the task covers; Macro-F1 on a 0-100 scale, higher is better.
Metric: Macro-F1 (%). Source: huggingface.co. Saturation forecast: Around February 2027. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 72.81 | #71 |
| 2 | Gemini 2.5 Pro | 69.74 | #159 |
| 3 | GPT-5 | 69.24 | #105 |
| 4 | GPT-5.1 (2025-11-13) | 66.61 | #107 |
| 5 | GPT-5.6 Luna | 63.51 | #79 |
| 6 | Claude Sonnet 4 (20250514) | 62.84 | #223 |
| 7 | Claude Sonnet 4 | 54.87 | #207 |
| 8 | Qwen 3.5 27B | 44.33 | #192 |
| 9 | Gemini 2.5 Flash | 43.36 | #255 |
| 10 | Qwen 3.6 27B | 40.89 | #182 |
| 11 | GPT-4.1 | 38.67 | #258 |
| 12 | Qwen 3.6 35B A3B | 28.99 | #238 |
| 13 | c4ai-command-a-03-2025 | 27.54 | #468 |
| 14 | Gemma 4 31B (IT) | 27.21 | #178 |
| 15 | Gemma 2 27B (IT) | 19 | #652 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sahara-pos-tagging · How It Works · Data refreshed daily, snapshot 2026-11-05.