I added a subscription to Ollama Cloud to my list of LLM providers. I was immediately disappointed to discover that Kimi K3 required an additional cash deposit, beyond the base $200 per year subscription, to even begin using that model. And from what it seems from my initial usage, Cline-Pass provides far more tokens over Ollama Cloud, even though Ollama's offering is 2.5x as expensive. Cline-Pass also has also made new cutting edge models available much more quickly than Ollama Cloud.

Ollama Cloud offers some benefits such as a zero data retention policy (although it's unfortunately not HIPAA compliant), and most important, they self-host all their models. To be clear, this means that they are not just a routing service such as OpenRouter, so if a disruption occurs at a downstream provider such as the hosted Deepseek API, that doesn't affect Ollama Cloud's availability. They operate totally separately, running the open source models on their own hardware infrastructure (owned and rented). For me, that's reason enough to keep an account with Ollama Cloud - I never want to be stuck without API access. Also, rates hikes by DeepSeek, as the minimum doubled expense we're currently seeing, don't affect Ollama.

Even with a stable of self-hosted models, I prefer to run LLM inference on an API, especially when models like Deepseek are basically not rate limited, and extremely cheap, like they are on Cline-Pass. That eliminates wear and tear, network use, and electricity consumption on my self-hosted machines.