That's what I was wondering as well. If I had to guess, the models that focus on tons of languages are often less good at say English->Spanish as a dedicated model, or a model that focuses on only a few high-resource languages.
Glad to see more offline-translation options though. Would be nice to have a benchmark for them soon to compare more easily.
Glad to see more offline-translation options though. Would be nice to have a benchmark for them soon to compare more easily.