Skip to content
Build Perch
Browse documentation

Ollama Cloud

Short answer: Ollama Cloud is your ollama.com subscription’s hosted models, connected as a third provider alongside OpenRouter and local Ollama. It’s a flat-rate way to run large open models like gpt-oss, DeepSeek, and Qwen without a local GPU.

What is Ollama Cloud?

Ollama Cloud is a separate, keyed subscription from Ollama (ollama.com, not your local install) that runs large open-weight models on Ollama’s own servers. Build Perch treats it as a distinct provider from local Ollama. Same underlying company, different tradeoffs: no local GPU needed, but billed by plan (Free/Pro/Max) rather than free-and-local.

Requests to Ollama Cloud are sent by the app itself using your saved key. Your key and your traffic go straight from your machine to ollama.com and never pass through any Build Perch server.

How do I set up Ollama Cloud?

  1. Open the Integrations window and select Ollama Cloud under LLM Providers.
  2. Create an API key at ollama.com/settings/keys and copy it.
  3. Paste the key into the Ollama Cloud row and save. Once saved, the key is never displayed again; it’s stored securely on your machine and used only to authenticate your requests to ollama.com.
  4. Open the model picker and select Ollama (Cloud) from the provider dropdown. Pick a model and connect.

Good to know

  • Billing is by plan. Ollama Cloud pricing shows as “Subscription” in the model picker rather than a $/token rate; usage counts against your plan’s limits, not a per-message charge.
  • Models are grouped by usage level (1–4), an approximate size proxy so free-tier users can find models likely to run on their plan: Level 1 is the lightest and most likely to work on a free subscription; Level 4 is the heaviest and almost certainly needs a paid plan. Level 1 is sorted first and its section is expanded by default; you still pick the model.
  • The catalog includes large open models (gpt-oss, DeepSeek, Qwen, Kimi, and others) enriched with real capability data (tool support, vision, context window) pulled live from each model.
  • Keyless local Ollama is unaffected. Ollama Cloud and local Ollama are deliberately separate providers with separate credentials; connecting one doesn’t touch the other, and you can run both side by side.
  • No media generation. Like local Ollama, Ollama Cloud serves chat models only; image, speech, music, and video generation are OpenRouter-only.
  • No OpenRouter-specific decorations. Responses stream token by token, but Ollama Cloud never carries OpenRouter-only extras like prompt-cache markers or the zero-data-retention flag, since neither applies to this provider.