List your company

Multimodal Reasoning on MathVista

89.8Pass@1Gemini 3-Pro
59.95267.70175.4583.199May 11, 2025Jul 2, 2025Aug 23, 2025Oct 15, 2025Dec 6, 2025Jan 27, 2026Mar 21, 2026

Evaluation results

MethodLinks
Gemini 3-Pro2026.0389.8
Seed1.82026.0387.7
Seed 1.5-VLthinking=true, decoding=greedy2025.0585.6
Seed1.5-VL Thinking2026.0385.6
Seed 1.5-VLthinking=false, decoding=greedy2025.0583
Gemini 1.5 Prothinking=true, decoding=greedy2025.0582.7
Gemini 2.5-Prosourced from technical report=true2026.0382.7
GPT-5.1 High2026.0380.6
Claude Sonnet-4.52026.0380.4
Qwen 2.5-VL 72Bthinking=false, decoding=greedy2025.0574.8
Claude 3.7 Sonnetthinking=true, decoding=sampling2025.0574.5
RAPBackbone=Qwen2.5-VL-7b, Sample=5,159, Time (h)=52.82025.0673.2
OpenAI o1thinking=true, decoding=greedy2025.0571.8
LIMRBackbone=Qwen2.5-VL-7b, Sample=8,136, Time (h)=1222025.0671.1
Qwen2.5-VL-7b-FullBackbone=Qwen2.5-VL-7b, Sample=54,931, Time (h)=93.22025.0670.7
LIMOBackbone=Qwen2.5-VL-7b, Sample=4,093, Time (h)=111.92025.0669.9
Qwen2.5-VL-7bBackbone=Qwen2.5-VL-7b2025.0668.7
s1Backbone=Qwen2.5-VL-7b, Sample=1,000, Time (h)=55.92025.0668.5
RAPBackbone=Qwen2.5-VL-3b, Sample=4,374, Time (h)=322025.0664.9
Qwen2.5-VL-3b-FullBackbone=Qwen2.5-VL-3b, Sample=54,931, Time (h)=46.52025.0664.5
GPT-4othinking=false, decoding=greedy2025.0563.8
Socratic-Solver-Geo (Stage3)Data Scale=2.5k, Curriculum Stage=32026.0263.55
GeoReasoningData Scale=10k2026.0263.4
LIMRBackbone=Qwen2.5-VL-3b, Sample=21,303, Time (h)=60.92025.0663.1
R-CoTData Scale=7.2k2026.0262.6
s1Backbone=Qwen2.5-VL-3b, Sample=1,000, Time (h)=38.42025.0662.6
KD (Our Synthesis)Data Scale=2.5k2026.0262.4
Socratic-Solver-Geo (Stage2)Data Scale=1k, Curriculum Stage=22026.0262.3
KD (Geo3K)Data Scale=3k2026.0262.1
Geo170k (G-LLaVA)Data Scale=10k2026.0262
TrustGeoGenData Scale=10k2026.0261.8
LIMOBackbone=Qwen2.5-VL-3b, Sample=2,679, Time (h)=120.52025.0661.8
PGPS9KData Scale=10k2026.0261.3
Qwen2.5-VL-3bBackbone=Qwen2.5-VL-3b2025.0661.3
Socratic-Solver-Geo (Stage1)Data Scale=0.4k, Curriculum Stage=12026.0261.2
Qwen2.5-VL-7B-InstructMode=Zero-shot2026.0261.1