SAHARA - Summarization: leaderboard
Summarization in African languages, scored against reference summaries, from the SAHARA African-language LLM leaderboard (UBC-NLP, ACL 2025). The task score is the mean over the languages the task covers; ROUGE-L on a 0-100 scale, higher is better.
Metric: ROUGE-L. Source: huggingface.co. Saturation forecast: Around 2034. 48 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4 | 22.7 | #207 |
| 2 | Gemini 3 Pro (Preview) | 21.89 | #71 |
| 3 | Llama 4 Scout Instruct | 19.45 | #547 |
| 4 | Qwen 3.6 27B | 19.29 | #182 |
| 5 | Qwen 3.6 35B A3B | 19.09 | #238 |
| 6 | Qwen 3.5 27B | 19.04 | #192 |
| 7 | Gemini 2.5 Pro | 17.68 | #159 |
| 8 | Llama 3.3 70B Instruct | 17.19 | #541 |
| 9 | Gemma 3 27B (IT) | 17.13 | #528 |
| 10 | Qwen 3.5 9B | 17.02 | #391 |
| 11 | Llama 3.1 70B Instruct | 16.92 | #567 |
| 12 | DeepSeek R1 Distill Llama 70B | 16.4 | #614 |
| 13 | GPT-4.1 | 16.3 | #258 |
| 14 | Qwen 3.5 4B | 16.22 | #514 |
| 15 | Gemma 3 12B (IT) | 16.16 | #672 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=sahara-summarization · How It Works · Data refreshed daily, snapshot 2026-11-05.