Pashto Evaluation of Audio AI Models Pashto Evaluation of Audio AI Models By Gooey.AI, CLEAR Global and the Gates Foundation • 28 Apr 2026 • Gooey.AI/language 0 5 10 15 20 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Accuracy Latency Lower is better More accurate GPT 5.5 — Accuracy 0.37, Latency 14.29GPT 5.5 GPT 5.5 + Omni — Accuracy 0.71, Latency 14.09GPT 5.5 + OmniMost Accurate Gemini 3.1 Pro — Accuracy 0.54, Latency 18.8Gemini 3.1 Pro Gemini 3.1 Pro + Omni — Accuracy 0.63, Latency 16.13Gemini 3.1 Pro + Omni Gemini 3.1 Flash-Lite — Accuracy 0.13, Latency 6.01Gemini 3.1 Flash-LiteFastest Gemini 3.1 Flash-Lite + Omni — Accuracy 0.53, Latency 8.04Gemini 3.1 Flash-Lite + Omni Kimi K2.6 + Omni — Accuracy 0.51, Latency 19.52Kimi K2.6 + Omni GPT-OSS 120B + Omni — Accuracy 0.44, Latency 10.72GPT-OSS 120B + Omni GPT-Realtime 1.5 — Accuracy 0.01, Latency 11.03GPT-Realtime 1.5 GPT-Realtime 1.5 + Omni — Accuracy 0.46, Latency 13.03GPT-Realtime 1.5 + Omni Gemma 4 26B + Omni — Accuracy 0.35, Latency 8.71Gemma 4 26B + Omni GPT 5.5 — Accuracy 0.37, Latency 14.290 GPT 5.5 + Omni — Accuracy 0.71, Latency 14.091 Gemini 3.1 Pro — Accuracy 0.54, Latency 18.82 Gemini 3.1 Pro + Omni — Accuracy 0.63, Latency 16.133 Gemini 3.1 Flash-Lite — Accuracy 0.13, Latency 6.014 Gemini 3.1 Flash-Lite + Omni — Accuracy 0.53, Latency 8.045 Kimi K2.6 + Omni — Accuracy 0.51, Latency 19.526 GPT-OSS 120B + Omni — Accuracy 0.44, Latency 10.727 GPT-Realtime 1.5 — Accuracy 0.01, Latency 11.038 GPT-Realtime 1.5 + Omni — Accuracy 0.46, Latency 13.039 Gemma 4 26B + Omni — Accuracy 0.35, Latency 8.7110 © 2026 Gooey.AI