Baidu's large vision-language MoE model — 424B total, 47B active.
Add a second model to compare.
This model hasn't been benchmarked yet.