List your company
ResearchFormerly WizWand

Multimodal Mathematical Reasoning on MathVerse mini

65.9AccuracyMetis
17.43630.01842.655.182Jul 9, 2025Sep 7, 2025Nov 6, 2025Jan 5, 2026Mar 6, 2026May 5, 2026Jul 5, 2026
Updated 3mo ago

Evaluation results

MethodLinks
MetisModel Category=Agentic Multimodal Models2026.0465.9
Qwen3-VL-8B-InstructModel Category=Open-source Models2026.0461.3
VL-Rethinker-7BModel Category=Text-only Reasoning Models2026.0454.2
D2IjusReasoning Strategy=D2I, Variant=jus, Backbone=Qwen2.5-VL-7B2025.0753.8
DeepEyesV2Model Category=Agentic Multimodal Models2026.0452.7
ThinkLite-VL-7BModel Category=Text-only Reasoning Models2026.0452.1
Qwen2.5-VL-7B w/ GRPO†Reasoning Mode=Intuitive, Backbone=Qwen2.5-VL-7B2025.0751.3
D2IlocReasoning Strategy=D2I, Variant=loc, Backbone=Qwen2.5-VL-7B2025.0751.1
Qwen2.5-VL-7B w/ GRPOReasoning Mode=Deliberate, Backbone=Qwen2.5-VL-7B2025.0750.6
D2IparReasoning Strategy=D2I, Variant=par, Backbone=Qwen2.5-VL-7B2025.0750.6
MM-Eureka-7BModel Category=Text-only Reasoning Models2026.0450.3
GPT-4o2025.0750.2
D2DlocReasoning Strategy=D2D, Variant=loc, Backbone=Qwen2.5-VL-7B2025.0749.1
D2DjusReasoning Strategy=D2D, Variant=jus, Backbone=Qwen2.5-VL-7B2025.0749.1
Qwen2.5-VL-7B*Backbone=Qwen2.5-VL-7B2025.0748.2
OpenVLThinker-7B2025.0747.9
D2DparReasoning Strategy=D2D, Variant=par, Backbone=Qwen2.5-VL-7B2025.0747.6
DeepEyesModel Category=Agentic Multimodal Models2026.0447.3
R1-Onevision-7B2025.0746.4
Qwen-2.5-VL-7B-InstructModel Category=Open-source Models2026.0445.6
PEPO_GBackbone=Qwen2.5-VL-3B-Instruct2026.0345.42
PEPO_DBackbone=InternVL3-2B-Instruct2026.0345.24
PEPO_GBackbone=InternVL3-2B-Instruct2026.0344.89
DAPOBackbone=Qwen2.5-VL-3B-Instruct2026.0344.44
PEPO_DBackbone=Qwen2.5-VL-3B-Instruct2026.0344.23
GRPOBackbone=InternVL3-2B-Instruct2026.0342.09
DAPOBackbone=InternVL3-2B-Instruct2026.0341.11
GRPOBackbone=Qwen2.5-VL-3B-Instruct2026.0340.54
High-Entropy RLBackbone=Qwen2.5-VL-3B-Instruct2026.0339.84
InternVL3-8BModel Category=Open-source Models2026.0439.8
InternVL2.5-8B2025.0739.5
GPT-4V2025.0739.4
BaseBackbone=Qwen2.5-VL-3B-Instruct, Mode=zero-shot2026.0337.66
InternVL2-8B2025.0737
CPOFine-tuning setting=CPO, Backbone=Qwen3-VL-2B2026.0736.12
GRPOFine-tuning setting=GRPO, Backbone=Qwen3-VL-2B2026.0732.51
GSPOFine-tuning setting=GSPO, Backbone=Qwen3-VL-2B2026.0732.16
Qwen2-VL-7B2025.0731.9
BaseFine-tuning setting=Base, Backbone=Qwen3-VL-2B2026.0731.12
LoRAFine-tuning setting=LoRA, Backbone=Qwen3-VL-2B2026.0726.32
BaseBackbone=InternVL3-2B-Instruct, Mode=zero-shot2026.0323.36
FFTFine-tuning setting=FFT, Backbone=Qwen3-VL-2B2026.0722.84
High-Entropy RLBackbone=InternVL3-2B-Instruct2026.0320.7
LLaVA-CoT-11B2025.0720.3
LLaVA-OneVisionModel Category=Open-source Models2026.0419.3