Share your thoughts, 1 month free Claude Pro on usSee more

Mathematical Problem Solving on MATH500

93Accuracy

Self-MoA

Updated 16d ago

Evaluation Results

Method
Self-MoA 2025.07	93	-	-	-	-	-
SMCS 2025.07	92.6	-	-	-	-	-
Symbolic-MoE* 2025.07	90.4	-	-	-	-	-
Self Consistency (Best on Validation) 2025.07	90.4	-	-	-	-	-
Majority Voting 2025.07	90.2	-	-	-	-	-
GLM-Z1-32B-0414 2025.07	90	-	-	-	-	-
Open-source Upper Bound 2025.07	90	-	-	-	-	-
RecursiveMAS 2026.04	88.2	-	893	-	2,320	-
Qwen3-32B 2025.07	88	-	-	-	-	-
Simple Router 2025.07	88	-	-	-	-	-
QwQ-32B 2025.07	87.8	-	-	-	-	-
MoA 2025.07	87.8	-	-	-	-	-
RecursiveMAS 2026.04	87.1	-	953	-	1,974	-
SelfBudgeter 2025.05	86.87	-	-	-	-	2,666.58
DeepSeek-R1-Distill-Qwen 2025.05	86.73	-	-	-	-	5,387.19
RecursiveMAS 2026.04	86.3	-	816	-	1,701	-
Recursive-TextMAS 2026.04	85.8	-	4,100	-	6,010	-
GPT-4.1 2025.07	85.8	-	-	-	-	-
TaH+ 2025.11	85.6	-	-	-	-	-
DeepSeek-R1-Distill-Qwen-32B 2025.07	85.6	-	-	-	-	-
TaH 2025.11	84.4	-	-	-	-	-
Recursive-TextMAS 2026.04	84.4	-	2,794	-	3,958	-
GPT-o3-mini 2025.07	84.4	-	-	-	-	-
EXAONE-Deep-32B 2025.07	84.38	-	-	-	-	-
Recursive-TextMAS 2026.04	84.2	-	1,471	-	2,401	-
Gemma-3-27b-it 2025.07	84	-	-	-	-	-
SoftThink 2025.11	83.2	-	-	-	-	-
Standard 2025.11	82.8	-	-	-	-	-
DeepSeek-R1-Distill-Llama-70B 2025.07	82.8	-	-	-	-	-
Qwen3-14B 2026.02	80.6	-	-	-	-	-
Eurus-2-7B-PRIME 2025.05	79.73	-	-	-	-	582.58
Qwen-2.5-72B-Instruct 2025.07	78.8	-	-	-	-	-
SelfBudgeter 2025.05	78.47	-	-	-	-	2,326.85
Qwen3-14B 2026.02	78.4	-	-	-	-	-
RecursiveMAS 2026.04	77.8	-	519	-	1,360	-
L1-Max 2025.05	76.73	-	-	-	-	1,753.42
RecursiveMAS 2026.04	76.6	-	495	-	1,096	-
Routing 2025.11	76.1	-	-	-	-	-
RecursiveMAS 2026.04	75.8	-	523	-	825	-
Qwen2.5-32b-Instruct 2025.07	75.6	-	-	-	-	-
Llama-3.3-Nemotron-Super-49B-v1 2025.07	75.2	-	-	-	-	-
DeepSeek-R1-Distill-Qwen 2025.05	74.93	-	-	-	-	5,327.12
GPT-4o 2025.07	74.6	-	-	-	-	-
TaH 2025.11	74.4	-	-	-	-	-
Claude-3.5-Sonnet 2025.07	74.2	-	-	-	-	-
Qwen2.5-Coder-32B-Instruct 2025.07	73.6	-	-	-	-	-
Claude-3.7-Sonnet 2025.07	73.2	-	-	-	-	-
TaH+ 2025.11	73	-	-	-	-	-
Llama-3.3-70B-Instruct 2025.07	73	-	-	-	-	-
HuatuoGPT-o1-72B 2025.07	73	-	-	-	-	-
Recursive-TextMAS 2026.04	72.5	-	2,117	-	2,204	-
E1-Math-1.5B 2025.05	72.47	-	-	-	-	2,088.44
Recursive-TextMAS 2026.04	71.9	-	1,185	-	1,368	-
TeleChat2-35B-32K 2025.07	70	-	-	-	-	-
Recursive-TextMAS 2026.04	69.1	-	3,059	-	2,952	-
SoftThink 2025.11	68.8	-	-	-	-	-
Standard 2025.11	68.4	-	-	-	-	-
AlwaysThink 2025.11	61.8	-	-	-	-	-
Qwen-2.5-7B-Simple-RL 2025.05	61.13	-	-	-	-	823.89
Routing 2025.11	60	-	-	-	-	-
InternLM2.5-20B-Chat 2025.07	55.2	-	-	-	-	-
TaH+ 2025.11	54.2	-	-	-	-	-
Llama3.1-70B 2026.02	53	-	-	-	-	-
Llama3.1-70B 2026.02	51.2	-	-	-	-	-
TaH 2025.11	51.2	-	-	-	-	-
SoftThink 2025.11	48.8	-	-	-	-	-
Standard 2025.11	47.2	-	-	-	-	-
ρ-EOS 2026.01	40.6	558.7	682.2	81.9	1,968	-
Fixed-Length Denoising 2026.01	40.4	585.8	1,024	57.2	3,663	-
Fixed-Length Denoising 2026.01	40	706.5	2,048	34.5	12,764	-
DAEDAL 2026.01	40	464.6	618.3	75.2	2,207	-
Fixed-Length Denoising 2026.01	38.4	423.9	512	82.8	1,202	-
ρ-EOS 2026.01	37	428.8	486.3	88.2	1,398	-
E1-Math-1.5B 2025.05	35.53	-	-	-	-	1,499.54
Fixed-Length Denoising 2026.01	35.4	244.7	256	95.6	462	-
Instruct Model + GIFT 2025.09	33	-	-	-	-	-
AlwaysThink 2025.11	32.8	-	-	-	-	-
Instruct Model + SFT 2025.09	31.8	-	-	-	-	-
Instruct Model + GIFT 2025.09	31.4	-	-	-	-	-
Instruct Model + SFT 2025.09	30	-	-	-	-	-
Fixed-Length Denoising 2026.01	29	123.3	128	96.3	208	-
Routing 2025.11	27.3	-	-	-	-	-
Fixed-Length Denoising 2026.01	24	61.7	64	96.4	98	-