Fig. 4: Performance comparison of proprietary MLLMs, open-source MLLMs on general multimodal benchmarks. | Nature Communications

Fig. 4: Performance comparison of proprietary MLLMs, open-source MLLMs on general multimodal benchmarks.

From: Efficient GPT-4V level multimodal large language model for deployment on edge devices

Fig. 4

MiniCPM-Llama-V 2.5, with only 8 billion parameters, outperforms leading open-source MLLMs and achieves superior results on the OpenCompass benchmark compared to proprietary models like GPT-4V-1106 and Gemini Pro. In addition, MiniCPM-V 2.0, with 2 billion parameters, significantly outperforms other MLLMs with fewer than 4 billion parameters.

Back to article page