← All guides · provider record →
Ollama Cloud — free tier setup
What you get free
Hosted big-model inference (~36 cloud models: DeepSeek V3.1 671B, Kimi K2 1T, Qwen3-Coder 480B, gpt-oss 120B...) on Ollama's own infrastructure, with an unusually strong data policy for a free tier: "Prompt or response data is never logged or trained on" (official). Per the catalog: "Ollama Cloud measures usage by GPU time, not tokens or requests. Free tier described as 'light usage' with session limits resetting every 5 hours and weekly limits every 7 days. Free plan: 1 cloud model at a time (concurrency)." The gotcha that matters most: this is the native Ollama API, NOT OpenAI-compatible — no /v1/chat/completions, no choices[]. Point OpenAI SDKs elsewhere or write the native shape.
Friction check
| Credit card | unknown (no card mention found on free plan pages) |
| Phone verification | unknown |
| Org/membership | none |
| Est. time to first key | ~5m (3 steps) |
| Trains on your prompts | no ("Prompt or response data is never logged or trained on") |
The rest of this guide — the key-obtain walkthrough, first request, the limits that will bite you, and the rotation runbook — ships with the Indie field library. What you’ve read above is the real first section, not a blurred preview; the remainder simply isn’t in this page.