List your company
ResearchFormerly WizWand

Mathematical Problem Solving on Gaokao MathQA

86.3AccuracyQwen2.5-Math-72B
13.532.451.370.2Feb 5, 2024Jun 13, 2024Oct 21, 2024Feb 27, 2025Jul 7, 2025Nov 13, 2025Mar 23, 2026
Updated 7mo ago

Evaluation results

MethodLinks
Qwen2.5-Math-72Bshot=4-shot, prompting=few-shot chain-of-thought2024.0986.3
TextGradBackbone=Qwen3-8B2026.0378.5
MemAPOBackbone=Qwen3-8B2026.0376.9
PromptBreederBackbone=Qwen3-8B2026.0376.6
ProTeGiBackbone=Qwen3-8B2026.0374.5
Qwen2-72Bshot=4-shot, prompting=few-shot chain-of-thought2024.0972.6
RaRBackbone=Qwen3-8B2026.0370.1
Qwen2-Math-72Bshot=4-shot, prompting=few-shot chain-of-thought2024.0969.5
Qwen2.5-Math-7Bshot=4-shot, prompting=few-shot chain-of-thought2024.0969.5
IOBackbone=Qwen3-8B2026.0368.5
CoTBackbone=Qwen3-8B2026.0365.7
Step-BackBackbone=Qwen3-8B2026.0364.5
OPROBackbone=Qwen3-8B2026.0363.3
MemAPOBackbone=GPT-4o-mini2026.0362.2
MemAPOBackbone=GPT-4o-mini2026.0361.8
TextGradBackbone=GPT-4o-mini2026.0359
Qwen2-Math-7Bshot=4-shot, prompting=few-shot chain-of-thought2024.0957.3
OPROBackbone=GPT-4o-mini2026.0356.2
TextGradBackbone=GPT-4o-mini2026.0356.2
Qwen2.5-Math-1.5Bshot=4-shot, prompting=few-shot chain-of-thought2024.0954.1
Qwen2-7Bshot=4-shot, prompting=few-shot chain-of-thought2024.0951.6
DeepSeek-Coder-V2-Lite-Baseshot=4-shot, prompting=few-shot chain-of-thought2024.0951.3
MathCoder-VL-8B#Params=8B2025.0551.2
SPOBackbone=Qwen3-8B2026.0351
Qwen2-Math-1.5Bshot=4-shot, prompting=few-shot chain-of-thought2024.0950.7
GPT-4-turbo#Params=-2025.0550
OPROBackbone=GPT-4o-mini2026.0349.4
SPOBackbone=GPT-4o-mini2026.0347
PromptBreederBackbone=GPT-4o-mini2026.0346.6
GPT-4V#Params=-2025.0545
ProTeGiBackbone=GPT-4o-mini2026.0345
ProTeGiBackbone=GPT-4o-mini2026.0345
Llama-3.1-70Bshot=4-shot, prompting=few-shot chain-of-thought2024.0943.3
StepBackBackbone=GPT-4o-mini2026.0341.8
Step-BackBackbone=GPT-4o-mini2026.0341.8
InternVL2-76B#Params=76B2025.0541.2
DeepSeekMath-Base-7Bshot=4-shot, prompting=few-shot chain-of-thought2024.0940.7
InternLM2-Math-Base-20Bshot=4-shot, prompting=few-shot chain-of-thought2024.0940.2
IOBackbone=GPT-4o-mini2026.0339.4
IOBackbone=GPT-4o-mini2026.0339.4
RephraseBackbone=GPT-4o-mini2026.0338.6
RaRBackbone=GPT-4o-mini2026.0338.6
MathCoder-VL-2B#Params=2B2025.0537.5
PromptBreederBackbone=GPT-4o-mini2026.0337.5
SPOBackbone=GPT-4o-mini2026.0335.9
Qwen2-1.5Bshot=4-shot, prompting=few-shot chain-of-thought2024.0935.6
CoTBackbone=GPT-4o-mini2026.0335.5
CoTBackbone=GPT-4o-mini2026.0335.5
DeepSeekMath-BaseSize=7B, Evaluation=Chain-of-thought prompting, Source=Open-Source Base Model2024.0235.3
Qwen-VL-Plus#Params=-2025.0533.8
InternVL2-26B#Params=26B2025.0533.4
InternVL2-8B#Params=8B2025.0532.5
Llama-3.1-8Bshot=4-shot, prompting=few-shot chain-of-thought2024.0928.5
LlemmaSize=34B, Evaluation=Chain-of-thought prompting, Source=Open-Source Base Model2024.0226.2
Qwen2-VL#Params=8B2025.0525
LlemmaSize=7B, Evaluation=Chain-of-thought prompting, Source=Open-Source Base Model2024.0223.6
MistralSize=7B, Evaluation=Chain-of-thought prompting, Source=Open-Source Base Model2024.0223.4
Deepseek-VL#Params=8B2025.0520
InternVL-Chat-2B-V1-5#Params=2B2025.0517.5
LLaVA-1.5-13B#Params=13B2025.0516.3