Molecular Optimization (MR) on TOMG-Bench benchmark leaderboard · SOTA2 Research

Molecular Optimization (MR) on TOMG-Bench

99.8Success Rate (SR)MolLingo

Evaluation Results

Method	Links
MolLingo2026.05		99.8	75.3	75.2
GPT-5.42026.05		90.3	69.1	62.4
Gemini-1.5-pro2026.05		78.76	67.44	53.1
GPT-4-turbo2026.05		73.88	68.21	50.4
OpenMolIns-xlarge2026.05		71.24	56.97	40.6
OpenMolIns-large2026.05		71.22	65.48	46.6
Claude-3.52026.05		69.62	71.12	49.5
GPT-4o2026.05		68.64	64.2	44
OpenMolIns-large2026.05		63.88	59.73	38.2
Claude-32026.05		60.94	63.98	39
REINVENT2026.03		59.5	11.6	6.9
OpenMolIns-medium2026.05		58.74	58.73	34.5
Llama3-70B-Instruct2026.05		56.84	60.32	34.3
Graph GA2026.03		50.9	12.2	6.2
MOLLEO2026.03		50.9	12.7	6.5
Bio-T5-base2026.03		50.6	16	8.1
BioT5-base2026.05		50.6	15.97	8.1
Mol-T5-large2026.03		45	10.7	4.8
MolT5-large2026.05		44.96	10.72	4.8
Llama3-8B-Instruct2026.05		43.32	47.93	20.8
Llama3.1-8B-Instruct2026.05		41.64	48.3	20.1
GPT-3.5-turbo2026.05		41.2	62.63	25.8
GPT-4o-mini2026.03		40.9	77.1	31.5
RePO2026.03		39.9	73.6	29.3
RePO2026.05		39.9	73.6	29.4
OpenMolIns-small2026.05		39.58	64.52	25.5
SFT2026.03		35.9	66.3	23.8
Chatglm-9B2026.05		35.14	68.2	24
OpenMolIns-light2026.05		35.08	64.35	22.6
GRPO2026.03		26.5	85	22.5
OpenMolIns-large2026.05		26.44	59.56	15.7
Baseline2026.03		25.2	68.5	17.3
MolT5-small2026.05		23.16	10.11	2.3
GRPO2026.03		21.4	88	18.8
Yi-1.5-9B2026.05		20.5	37.24	7.6
Mistral-7B-Instruct-v0.22026.05		19.08	25.78	4.9
MolT5-base2026.05		18.56	10.73	2
Llama-3.2-1B-Instruct2026.05		8.22	44.1	3.6
Qwen2-7B-Instruct2026.05		0.02	41.23	0