Multi-step Scientific Tool-use on SciAgentBench Main Results 1.0 benchmark leaderboard · SOTA2 Research

Multi-step Scientific Tool-use on SciAgentBench Main Results 1.0

32.3SR (w/o Tools)GPT-5

Evaluation Results

Method	Links
GPT-52026.02		32.3	41.3	9	0.24	46.3	43.8	28.6	32.3	60.6	38.6	30.9
Grok-4-12026.02		30.4	40.3	9.9	0.25	47.2	38.2	32.4	30	52.8	43.2	27.6
Gemini-2.5-Pro-Think2026.02		28.9	28.8	-0.1	0.19	33.3	28.9	21.2	21.9	51.7	28.9	11.7
Gemini-2.5-Flash2026.02		28.5	32.7	4.2	0.21	38.3	32.4	28.6	17.2	49.5	32.7	22
O4-mini2026.02		27.8	31.1	3.3	0.24	31.2	35.5	30.6	20	53.3	35.2	9.7
O32026.02		26.6	32	5.4	0.26	35.5	37.3	32.4	6.5	50.4	31.1	20.2
GLM-4.6V2026.02		26	30.9	4.9	0.25	30.9	37.5	22.2	18.8	44.4	31.4	21.6
Gemini-2.5-Pro2026.02		24.8	32.6	7.8	0.21	37.3	35.1	26.5	18.8	56.5	33.4	16.1
Qwen3-VL-235B-Think2026.02		24.4	28	3.6	0.16	30.6	29.5	22.9	22.6	53.9	26.9	15.1
Qwen3-VL-32B-Think2026.02		24.4	27.9	3.5	0.17	33	31.2	8.8	22.6	49.1	26.2	14.6
SciAgent-8B2026.02		23.3	30.1	6.8	0.16	33	35.2	9.1	31	41.9	27.5	19.6
Qwen3-VL-235B-Inst2026.02		23	23.9	0.9	0.16	28.1	26.5	5	17.2	50	21.8	2.4
Qwen3-VL-32B-Inst2026.02		22.8	27.4	4.6	0.15	31.8	29.3	20	16.1	47.1	25.4	15.1
Claude-4-Sonnet2026.02		22.4	35.9	13.5	0.19	39.4	39.5	27	25	55.6	36.4	16.9
Qwen3-VL-8B-Inst2026.02		18.4	23.4	5	0.09	24	28.6	7.1	24.1	38.7	20.9	8.2
SciAgent-4B2026.02		17.4	25.2	7.8	0.13	28.4	28.4	14.7	19.4	43.3	27.5	13.5
GPT-4o2026.02		17.1	18.7	1.6	0.14	21.3	20.5	8.6	16	27.3	16.4	7.5
Qwen3-VL-4B-Inst2026.02		17	19.7	2.7	0.1	23.8	20.6	10.3	13.3	40.7	15.2	10
Pixtral-12B2026.02		7.8	7.2	-0.6	0.07	7.5	6.3	5.9	10	12.1	2.9	3.7