Multimodal Mathematical Reasoning on MathVista mini (test)
77.1Overall AccuracyReLaX-VL-7B
Evaluation results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ReLaX-VL-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 77.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SRPO-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 75.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| VL-Rethinker-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 74.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-72BModel Category=General Multimodal LLM, Parameters=72B2025.12 | 74.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFT-MModel=Qwen2.5-VL-7B, Paradigm=SFT-M2026.02 | 74.07 | — | — | — | — | — | — | — | — | — | — | — | — | |
| OpenAI o1Category=Close-Source MLLMs2025.07 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| OpenAI o12025.07 | 73.9 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Llama 4 MaverickCategory=Close-Source MLLMs2025.07 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Llama 4 Maverick2025.07 | 73.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFT-RSModel=Qwen2.5-VL-7B, Paradigm=SFT-RS2026.02 | 73.52 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Vision-R1-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 73.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini-2.0-Flash2026.02 | 73.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| MM-Eureka-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 73 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFTModel=Qwen2.5-VL-7B, Paradigm=SFT2026.02 | 71.67 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ThinkOmniBase Model=Omni-R1-7B, LRM Guide=Qwen32026.02 | 71.3 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ReLaX-VL-3BModel Category=Reasoning Multimodal LLM, Parameters=3B2025.12 | 70.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-DPPost-training=Disc./Preference Reward2025.07 | 70.4 | 72.8 | 72.6 | 77.2 | 68.5 | 54.5 | 71.1 | 69.6 | 69.3 | 27 | 43.1 | 66.9 | 77.2 | |
| Qwen2.5-VL-DPBackbone=Qwen2.5-VL2025.07 | 70.4 | 72.8 | 72.6 | 77.2 | 68.5 | 54.5 | 71.1 | 69.6 | 69.3 | 27 | 43.1 | 66.9 | 77.2 | |
| OpenVLThinker-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ThinkOmniBase Model=Qwen2.5-Omni-7B, LRM Guide=Qwen32026.02 | 70.2 | — | — | — | — | — | — | — | — | — | — | — | — | |
| GRPOModel=Qwen2.5-VL-7B, Paradigm=GRPO2026.02 | 69.81 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFT-MModel=Qwen2.5-VL-3B, Paradigm=SFT-M2026.02 | 68.89 | — | — | — | — | — | — | — | — | — | — | — | — | |
| HumanOmniV2Notes=Evaluated using authors' scripts2026.02 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ThinkOmniBase Model=Qwen2.5-Omni-7B, LRM Guide=DeepSeek2026.02 | 68.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Kimi-VL-16BModel Category=General Multimodal LLM, Parameters=16B2025.12 | 68.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Ola2026.02 | 68.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| SFTModel=Qwen2.5-VL-3B, Paradigm=SFT2026.02 | 68.33 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-7BModel Category=General Multimodal LLM, Parameters=7B2025.12 | 68.2 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-7BCategory=Open-Source MLLMs2025.07 | 68.2 | 72.5 | 66.8 | 76.9 | 66.7 | 54.3 | 70.1 | 68.7 | 66.9 | 26.9 | 43 | 65.7 | 76.1 | |
| Qwen2.5-VL-7BParameters=7B2025.07 | 68.2 | 72.5 | 66.8 | 76.9 | 66.7 | 54.3 | 70.1 | 68.7 | 66.9 | 26.9 | 43 | 65.7 | 76.1 | |
| VReST-VoteBackbone=Qwen2.5-VL-3B-Instruct, Search Strategy=MCTS + Voting2025.06 | 67.4 | 72.12 | 72.12 | 67.2 | 63.92 | 58.1 | 65.48 | 60.34 | 69.04 | 40.54 | 48.61 | 72.13 | 74.75 | |
| MM-Eureka-8BModel Category=Reasoning Multimodal LLM, Parameters=8B2025.12 | 67.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-Omni-7B2026.02 | 66.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| VReSTBackbone=Qwen2.5-VL-3B-Instruct, Search Strategy=MCTS2025.06 | 66.5 | 74.35 | 71.15 | 67.2 | 58.86 | 55.31 | 64.41 | 59.21 | 67.78 | 37.84 | 45.14 | 66.39 | 77.08 | |
| GRPOModel=Qwen2.5-VL-3B, Paradigm=GRPO2026.02 | 66.48 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ThinkOmniBase Model=Omni-R1-7B, LRM Guide=DeepSeek2026.02 | 66.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ToTBackbone=Qwen2.5-VL-3B-Instruct, Search Strategy=Tree of Thoughts2025.06 | 65.6 | 69.14 | 70.19 | 67.74 | 60.13 | 57.54 | 63.35 | 58.64 | 67.36 | 24.32 | 47.22 | 68.03 | 74.42 | |
| SFT-RSModel=Qwen2.5-VL-3B, Paradigm=SFT-RS2026.02 | 65.19 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Omni-R1Notes=Evaluated using authors' scripts2026.02 | 64.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Omni-R12026.02 | 64.7 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Intern2.5-VL-8BModel Category=General Multimodal LLM, Parameters=8B2025.12 | 64.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| CoT-VoteBackbone=Qwen2.5-VL-3B-Instruct, Search Strategy=Voting2025.06 | 64.2 | 69.89 | 67.79 | 65.05 | 58.86 | 55.31 | 60.5 | 57.51 | 65.27 | 29.73 | 43.75 | 68.85 | 74.42 | |
| GPT-4o2026.02 | 63.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4oCategory=Close-Source MLLMs2025.07 | 63.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4o2025.07 | 63.8 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Baichuan-Omni-1.52026.02 | 63.6 | — | — | — | — | — | — | — | — | — | — | — | — | |
| R1-VL-7BModel Category=Reasoning Multimodal LLM, Parameters=7B2025.12 | 63.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| CantorBackbone=Qwen2.5-VL-3B-Instruct2025.06 | 63.3 | 69.14 | 68.75 | 61.29 | 58.86 | 54.19 | 61.21 | 55.24 | 66.53 | 24.32 | 43.75 | 66.39 | 73.75 | |
| Llava-OV-7BModel Category=General Multimodal LLM, Parameters=7B2025.12 | 63.2 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Mulberry-7BCategory=Open-Source MLLMs2025.07 | 63.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Mulberry-7BParameters=7B2025.07 | 63.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-VL-3BModel Category=General Multimodal LLM, Parameters=3B2025.12 | 62.3 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Best-of-NBackbone=Qwen2.5-VL-3B-Instruct, Search Strategy=Best-of-N2025.06 | 62 | 69.52 | 66.35 | 60.75 | 55.06 | 53.07 | 58.01 | 54.39 | 65.27 | 29.73 | 41.67 | 64.75 | 72.43 | |
| Qwen2-VL-DPPost-training=Disc./Preference Reward2025.07 | 60.9 | 70.7 | 56.4 | 69.8 | 64.6 | 48.7 | 50.9 | 60.4 | 47.2 | 25.4 | 40.3 | 65.6 | 71.1 | |
| Qwen2-VL-DPBackbone=Qwen2-VL2025.07 | 60.9 | 70.7 | 56.4 | 69.8 | 64.6 | 48.7 | 50.9 | 60.4 | 47.2 | 25.4 | 40.3 | 65.6 | 71.1 | |
| HumanCategory=Heuristics Baselines2025.07 | 60.3 | 59.7 | 48.4 | 73 | 63.2 | 55.9 | 50.9 | 59.2 | 51.4 | 40.7 | 53.8 | 64.9 | 63.9 | |
| Human2025.07 | 60.3 | 59.7 | 48.4 | 73 | 63.2 | 55.9 | 50.9 | 59.2 | 51.4 | 40.7 | 53.8 | 64.9 | 63.9 | |
| QABackbone=Qwen2.5-VL-3B-Instruct2025.06 | 59.9 | 66.91 | 64.42 | 58.6 | 53.16 | 51.4 | 56.23 | 52.12 | 61.92 | 21.62 | 39.58 | 62.3 | 71.1 | |
| CoTBackbone=Qwen2.5-VL-3B-Instruct2025.06 | 58.7 | 65.43 | 63.46 | 56.99 | 51.9 | 50.84 | 55.52 | 50.99 | 60.67 | 18.92 | 39.58 | 60.66 | 69.44 | |
| Intern2-VL-8BModel Category=General Multimodal LLM, Parameters=8B2025.12 | 58.3 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2-VL-7BModel Category=General Multimodal LLM, Parameters=7B2025.12 | 58.2 | — | — | — | — | — | — | — | — | — | — | — | — | |
| ThinkOmniBase Model=Qwen2.5-Omni-3B, LRM Guide=Qwen32026.02 | 58.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2-VL-7BCategory=Open-Source MLLMs2025.07 | 57.6 | 65.1 | 41.8 | 66.1 | 60.1 | 53.7 | 44.5 | 56.4 | 43.1 | 24.3 | 39.6 | 63.1 | 69.4 | |
| Qwen2-VL-7BParameters=7B2025.07 | 57.6 | 65.1 | 41.8 | 66.1 | 60.1 | 53.7 | 44.5 | 56.4 | 43.1 | 24.3 | 39.6 | 63.1 | 69.4 | |
| ThinkOmniBase Model=Qwen2.5-Omni-3B, LRM Guide=DeepSeek2026.02 | 56.1 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen2.5-Omni-3B2026.02 | 56 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Multimath-7BCategory=Open-Source MLLMs2025.07 | 50 | — | 66.8 | 61.8 | — | — | — | — | — | — | — | — | — | |
| Multimath-7BParameters=7B2025.07 | 50 | — | 66.8 | 61.8 | — | — | — | — | — | — | — | — | — | |
| GPT-4VCategory=Close-Source MLLMs2025.07 | 49.9 | 43.1 | 50.5 | 57.5 | 65.2 | 38 | 53 | 49 | 51 | 21.6 | 20.1 | 63.1 | 55.8 | |
| GPT-4V2025.07 | 49.9 | 43.1 | 50.5 | 57.5 | 65.2 | 38 | 53 | 49 | 51 | 21.6 | 20.1 | 63.1 | 55.8 | |
| Math-PUMA-7BCategory=Open-Source MLLMs2025.07 | 47.9 | — | 48.1 | — | — | — | — | — | 47.3 | — | — | — | — | |
| Math-PUMA-7BParameters=7B2025.07 | 47.9 | — | 48.1 | — | — | — | — | — | 47.3 | — | — | — | — | |
| Math-LLaVACategory=Open-Source MLLMs2025.07 | 46.6 | 37.2 | 57.7 | 56.5 | 51.3 | 33.5 | 53 | 40.2 | 56.5 | 16.2 | 33.3 | 49.2 | 43.9 | |
| Math-LLaVA2025.07 | 46.6 | 37.2 | 57.7 | 56.5 | 51.3 | 33.5 | 53 | 40.2 | 56.5 | 16.2 | 33.3 | 49.2 | 43.9 | |
| LLaVA-Next-34BCategory=Open-Source MLLMs2025.07 | 46.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| LLaVA-Next-34BParameters=34B2025.07 | 46.5 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Claude 3 HaikuCategory=Close-Source MLLMs2025.07 | 46.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Claude 3 Haiku2025.07 | 46.4 | — | — | — | — | — | — | — | — | — | — | — | — | |
| Gemini 1.0 ProCategory=Close-Source MLLMs2025.07 | 45.2 | 47.6 | 40.4 | 39.2 | 61.4 | 39.1 | 45.2 | 38.8 | 41 | 10.8 | 32.6 | 54.9 | 56.8 | |
| Gemini 1.0 ProModel Type=Pro2025.07 | 45.2 | 47.6 | 40.4 | 39.2 | 61.4 | 39.1 | 45.2 | 38.8 | 41 | 10.8 | 32.6 | 54.9 | 56.8 | |
| Qwen-VL-PlusCategory=Close-Source MLLMs2025.07 | 43.3 | 54.6 | 38.5 | 31.2 | 55.1 | 34.1 | 39.1 | 32 | 39.3 | 18.9 | 26.4 | 59 | 56.1 | |
| Qwen-VL-Plus2025.07 | 43.3 | 54.6 | 38.5 | 31.2 | 55.1 | 34.1 | 39.1 | 32 | 39.3 | 18.9 | 26.4 | 59 | 56.1 | |
| LLaVA-1.5-DPPost-training=Disc./Preference Reward2025.07 | 42.2 | 32.3 | 56.2 | 51.6 | 45.6 | 39.7 | 43.8 | 41.6 | 46 | 15.4 | 38.9 | 46.7 | 40.9 | |
| LLaVA-1.5-DPBackbone=LLaVA-1.52025.07 | 42.2 | 32.3 | 56.2 | 51.6 | 45.6 | 39.7 | 43.8 | 41.6 | 46 | 15.4 | 38.9 | 46.7 | 40.9 | |
| MiniCPM-V2Category=Open-Source MLLMs2025.07 | 40.6 | 53.2 | 26 | 37.1 | 44.3 | 39.1 | 28.5 | 33.1 | 28 | 10.8 | 39.6 | 48.4 | 51.8 | |
| MiniCPM-V2Version=V22025.07 | 40.6 | 53.2 | 26 | 37.1 | 44.3 | 39.1 | 28.5 | 33.1 | 28 | 10.8 | 39.6 | 48.4 | 51.8 | |
| SPHINX-V2-13BCategory=Open-Source MLLMs2025.07 | 36.7 | 54.6 | 16.4 | 23.1 | 41.8 | 43 | 20.6 | 33.4 | 17.6 | 24.3 | 21.5 | 43.4 | 51.5 | |
| SPHINX-V2-13BVersion=V2, Parameters=13B2025.07 | 36.7 | 54.6 | 16.4 | 23.1 | 41.8 | 43 | 20.6 | 33.4 | 17.6 | 24.3 | 21.5 | 43.4 | 51.5 | |
| OmniLMM-12BCategory=Open-Source MLLMs2025.07 | 34.9 | 45 | 17.8 | 26.9 | 44.9 | 39.1 | 23.1 | 32.3 | 20.9 | 18.9 | 27.8 | 45.9 | 44.2 | |
| OmniLMM-12BParameters=12B2025.07 | 34.9 | 45 | 17.8 | 26.9 | 44.9 | 39.1 | 23.1 | 32.3 | 20.9 | 18.9 | 27.8 | 45.9 | 44.2 | |
| Gemini 1.0 Nano 2Category=Close-Source MLLMs2025.07 | 30.6 | 28.6 | 23.6 | 30.6 | 41.8 | 31.8 | 27.1 | 29.8 | 26.8 | 10.8 | 20.8 | 40.2 | 33.5 | |
| Gemini 1.0 Nano 2Model Size=Nano 22025.07 | 30.6 | 28.6 | 23.6 | 30.6 | 41.8 | 31.8 | 27.1 | 29.8 | 26.8 | 10.8 | 20.8 | 40.2 | 33.5 | |
| LLaVA-1.5-13BCategory=Open-Source MLLMs2025.07 | 27.7 | 23.8 | 22.7 | 18.3 | 40.5 | 30.2 | 25.3 | 26.4 | 22.8 | 21.6 | 26.4 | 35.3 | 23.6 | |
| LLaVA-1.5-13BVersion=1.5, Parameters=13B2025.07 | 27.7 | 23.8 | 22.7 | 18.3 | 40.5 | 30.2 | 25.3 | 26.4 | 22.8 | 21.6 | 26.4 | 35.3 | 23.6 | |
| SPHINX-V1-13BCategory=Open-Source MLLMs2025.07 | 27.5 | 23.4 | 23.1 | 21.5 | 39.9 | 34.1 | 25.6 | 28.1 | 23.4 | 16.2 | 17.4 | 40.2 | 23.6 | |
| SPHINX-V1-13BVersion=V1, Parameters=13B2025.07 | 27.5 | 23.4 | 23.1 | 21.5 | 39.9 | 34.1 | 25.6 | 28.1 | 23.4 | 16.2 | 17.4 | 40.2 | 23.6 | |
| Frequent GuessCategory=Heuristics Baselines2025.07 | 26.3 | 22.7 | 34.1 | 20.4 | 31 | 24.6 | 33.1 | 18.7 | 31.4 | 24.3 | 19.4 | 32 | 20.9 | |
| Frequent Guess2025.07 | 26.3 | 22.7 | 34.1 | 20.4 | 31 | 24.6 | 33.1 | 18.7 | 31.4 | 24.3 | 19.4 | 32 | 20.9 | |
| LLaVA-13BCategory=Open-Source MLLMs2025.07 | 26.1 | 26.8 | 29.3 | 16.1 | 32.3 | 26.3 | 27.3 | 20.1 | 28.8 | 24.3 | 18.3 | 37.3 | 25.1 | |
| LLaVA-13BParameters=13B2025.07 | 26.1 | 26.8 | 29.3 | 16.1 | 32.3 | 26.3 | 27.3 | 20.1 | 28.8 | 24.3 | 18.3 | 37.3 | 25.1 |