Right now, there is a whirling gyre around what to look at next within the LLM world. Should a developer be playing with existing models, extending existing models, or creating those models from scratch? After updating my post about using local models with Ollama, I thought I’d switch to looking at a cloud-based framework for this post.
Replicate is a cloud platform that allows users to easily run and fine-tune various open source AI models. They claim that about 20,000 open source models are available, including (inevitably) DeepSeek. But…