Kinyarwanda Evaluation of Audio AI Models Kinyarwanda Evaluation of Audio AI Models By Gooey.AI, CLEAR Global and the Gates Foundation • 30 Aug 2026 • Gooey.AI/language 0 2 4 6 8 10 12 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Accuracy Latency Lower is better More accurate Gemini 3.1 Pro — Accuracy 0.9, Latency 8.88Gemini 3.1 Pro Gemini 3.1 Pro + Omni — Accuracy 0.86, Latency 9.48Gemini 3.1 Pro + Omni Gemini 3.1 Pro + Intron — Accuracy 0.86, Latency 10.7Gemini 3.1 Pro + Intron GPT 5.6 Sol + Omni — Accuracy 0.85, Latency 6.94GPT 5.6 Sol + Omni GPT 5.6 Sol + Intron — Accuracy 0.81, Latency 8.65GPT 5.6 Sol + Intron Claude Sonnet 5 + Omni — Accuracy 0.82, Latency 7.11Claude Sonnet 5 + Omni Claude Sonnet 5 + Intron — Accuracy 0.87, Latency 8.85Claude Sonnet 5 + Intron Kimi K3 + Omni — Accuracy 0.78, Latency 6.71Kimi K3 + OmniFastest Kimi K3 + Intron — Accuracy 0.68, Latency 8.47Kimi K3 + Intron DeepSeek V4 + Omni — Accuracy 0.92, Latency 7.14DeepSeek V4 + OmniMost Accurate DeepSeek V4 + Intron — Accuracy 0.85, Latency 8.53DeepSeek V4 + Intron Gemini 3.1 Pro — Accuracy 0.9, Latency 8.880 Gemini 3.1 Pro + Omni — Accuracy 0.86, Latency 9.481 Gemini 3.1 Pro + Intron — Accuracy 0.86, Latency 10.72 GPT 5.6 Sol + Omni — Accuracy 0.85, Latency 6.943 GPT 5.6 Sol + Intron — Accuracy 0.81, Latency 8.654 Claude Sonnet 5 + Omni — Accuracy 0.82, Latency 7.115 Claude Sonnet 5 + Intron — Accuracy 0.87, Latency 8.856 Kimi K3 + Omni — Accuracy 0.78, Latency 6.717 Kimi K3 + Intron — Accuracy 0.68, Latency 8.478 DeepSeek V4 + Omni — Accuracy 0.92, Latency 7.149 DeepSeek V4 + Intron — Accuracy 0.85, Latency 8.5310 © 2026 Gooey.AI