← All models

DeepSeek-VL2-Small

DeepSeek · 2024-12-13 · 16.1B parameters

DeepSeek-VL2-Small is the mid-size model of DeepSeek-VL2, released December 13, 2024: a SigLIP vision encoder with dynamic tiling feeding a DeepSeekMoE language model with MLA (27 layers, 64 routed and 2 shared experts, 6 routed active). Its released weights contain 16,148,349,504 parameters, about 2.8B activated per token by DeepSeek's rounded figure, with a 4,096-token context.

Benchmark scores