Dear readers, if you could spare a small tip, consider it as treating me to a soda. If not, that’s totally fine—no need for any negativity.
[pay amount=6]
In mid-March 2026, Ollama partnered with NVIDIA Cloud Provider (NCP) to host open models, which are only suitable for light usage. The free tier allows you to run only one cloud model at a time, with a quota of about 500,000 tokens every 5 hours and about 1 million tokens every 7 days.
Visit
Register for an account
Then go to this page to add your API key. Currently, Gemma4, Qwen3.5, Kimi K2.5, GLM-5, GPT-OSS, and MiniMax M2.7 are supported (only the models in the list with names ending in -cloud are available).
Thanks for reading!
[/pay]