Retrieval on Semantic Benchmarks benchmark leaderboard · SOTA2 Research

Retrieval on Semantic Benchmarks

100Success RateQwen3C-30B-Inst-Q3_K_M

Evaluation Results

Method	Links
Qwen3C-30B-Inst-Q3_K_M2026.04		100	80	62.7	58	66	74	78	80	22.2	1.42
qwen3-coder:30b2026.04		100	80	51.8	46	52	72	76	80	21.1	1.34
BM252026.04		100	78	54.1	48	60	66	78	78	—	—
deepseek-coder-v3:16b2026.04		100	66	39.6	32	46	52	60	66	47.9	1.54
Semantic2026.04		100	54	27.9	22	32	38	52	54	—	—
Simple (bag-of-words)2026.04		100	54	27.9	22	32	38	52	54	—	—
qwen2.5-coder:7b2026.04		100	54	31.1	26	36	40	52	54	59.3	0.84
Qwen3C-30B-Inst-Q3_K_M2026.04		100	42	34.8	32	38	38	40	42	22.7	1.4
deepseek-coder:33b2026.04		100	12	8.7	8	8	12	12	12	180.4	0.14
granite-code:34b2026.04		100	2	1	0	2	2	2	2	111.3	0.04
Hybrid (BM25+semantic)2026.04		100	2	0.1	0	0	0	0	2	—	—
qwen2.5-coder:1.5b2026.04		100	0	0	0	0	0	0	0	63.7	0
qwen2.5-coder:3b2026.04		94	60	36.6	28	45	53	55	57	210.4	1.51
qwen2.5-coder:14b2026.04		82	56	46.1	41	51	51	56	56	73.4	1.05
microcoder-oss-20b2026.04		54	0	0	0	0	0	0	0	47.6	0
microcoder-deepseekr1-14.82026.04		52	73	45.3	38	46	65	69	73	107.4	4.15
qwen3vl-4b-orlex2026.04		12	8	8	8	8	8	8	8	37.9	0.14