Speech-to-Speech reasoning on Kimi-Audio-Evalkit s2s benchmark leaderboard · SOTA2 Research

Speech-to-Speech reasoning on Kimi-Audio-Evalkit s2s

78Llama Q ScoreGemini 2.5-Flash

Evaluation Results

Method	Links
Gemini 2.5-Flash2026.06		78	67	78	3.3	3.58	2.31	74	4.82	72.2
FlexiSLM-7B-Stage22026.06		75	59	59	3.43	3.35	2.24	57	4.93	66.1
FlexiSLM-7B-Stage32026.06		75	60	60	3.5	3.48	2.33	60	4.83	67.2
Fun-Audio-Chat-8B2026.06		74	60	59	3.77	3.46	2.55	54	4.81	67.3
FlexiSLM-7B-Stage22026.06		74	58	58	3.29	3.29	2.21	57	4.9	65.1
FlexiSLM-7B-Stage32026.06		74	59	58	3.44	3.42	2.26	59	4.87	66.2
FlexiSLM-7B-Stage32026.06		73	56	58	3.38	3.32	2.28	59	4.85	65.3
Qwen3-Omni-30B-A3B2026.06		72	63	67	3.6	3.58	2.6	62	4.89	69.7
Qwen2.5-Omni-7B2026.06		72	59	57	3.07	3.24	2.12	52	4.88	63.3
Mimo-Audio-7B2026.06		71	55	48	2.9	3.06	2.12	47	4.43	59
FlexiSLM-7B-Stage22026.06		71	56	56	3.23	3.25	2.12	55	4.92	63.6
FlexiSLM-7B-Stage32026.06		71	55	57	3.31	3.3	2.21	58	4.85	64.3
FlexiSLM-7B-Stage22026.06		70	59	57	3.33	3.29	2.14	54	4.86	64.1
Kimi-Audio-7B2026.06		69	56	54	2.74	2.46	2.11	37	4.73	57.2
FlexiSLM-7B-Stage32026.06		67	52	52	3.05	3.04	2.09	53	4.79	60.4
FlexiSLM-7B-Stage32026.06		64	48	48	2.84	2.83	1.85	48	4.67	56.5
GLM-4-Voice-9B2026.06		62	48	47	2.82	2.69	1.97	45	4.49	55.2