TukaBench — leaderboard
Metric: ASR â African Languages (self-reported). Source: benchmarklist.com. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-3.5 | 38.1 |
| 2 | Grok 3 | 31.5 |
| 3 | GPT-4o | 23.5 |
| 4 | Llama 4 Maverick | 21.9 |
| 5 | DeepSeek V3.2 | 18.5 |
| 6 | Gemma 4 31B | 16.6 |
| 7 | Gemma 3 27B | 15.6 |
| 8 | Grok 4.3 | 11.6 |
| 9 | GPT-OSS-120B | 10.7 |
| 10 | GPT-5.2 | 9.3 |
| 11 | Grok 4 | 9 |
| 12 | Qwen 3.5 27B | 7.9 |
| 13 | Gemini 3.1 Pro (Preview) | 2.4 |
Interactive version: theaggregate.ai/benchmark?slug=tukabench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.