Ollama Cloud
Cloud-hosted open models at NATIVE weights (GLM 5.2 1M-ctx, Kimi K2.7 Code, DeepSeek V4 Pro) with a free tier in the $0 plan; Pro $20/mo (50x usage, 3 concurrent), Max $100/mo (5x Pro, 10 concurrent); zero data retention.
- Card required
- No
Ollama Cloud scores 7.4/10 for free value in LLM. The main constraint: Free $0 plan includes access to cloud models with reported 'light usage' limits: session limits reset every 5 hours and weekly limits every 7 days; usage measured by GPU time, not tokens. Pro ($20/mo) and Max ($100/mo) raise limits 50x/250x.
Free value score
7.4 / 10Show the math
Key metrics
Models
7 free · 20 totalLimits may be stale — last re-verified outside this section's freshness window.
| Model | Family | AA | Access |
|---|---|---|---|
| glm-4.7 | GLM | 33.8 | free |
| gemma4 | Gemma | — | free |
| gpt-oss | GPT | — | free |
| minimax-m2.1 | MiniMax | — | free |
| minimax-m2.5 | MiniMax | — | free |
| nemotron-3-super | Nemotron | — | free |
| qwen3-coder | Qwen | — | free |
| glm-5.2 | GLM | 51.1 | paid |
| minimax-m3 | MiniMax | 44.4 | paid |
| deepseek-v4-pro | DeepSeek | 44.3 | paid |
| kimi-k2.6 | Kimi | 42.8 | paid |
| kimi-k2.7-code | Kimi | 42 | paid |
| deepseek-v4-flash | DeepSeek | 40.3 | paid |
| glm-5.1 | GLM | 40.2 | paid |
| glm-5 | GLM | 39.5 | paid |
| kimi-k2.5 | Kimi | 38.1 | paid |
| minimax-m2.7 | MiniMax | 38.1 | paid |
| nemotron-3-ultra | Nemotron | 37.8 | paid |
| gemini-3-flash-preview | Gemini | 27.4 | paid |
| qwen3.5 | Qwen | — | paid |
What you get
Cloud model access included in the free $0 plan
Same CLI/API as local Ollama
Verified on official Ollama pricing page
Time-windowed limits (5h / 7d), not a one-time trial
Gotchas
- 1 concurrent cloud model (Free)
Free plan allows only one cloud model running at a time; additional requests are queued and rejected if the queue is full.
Sources
-
"Each plan has session limits that reset every 5 hours and weekly limits that reset every 7 days."
https://ollama.com/pricing -
"Usage reflects actual utilization of Ollama's cloud infrastructure - primarily GPU time, which depends on model size and request duration. ... Ollama doesn't cap you at a set number of tokens."
https://ollama.com/pricing -
"qwen3-coder:480b-cloud | Available in Free Account ... gpt-oss:120b-cloud | Available in Free Account ... minimax-m3:cloud | Available in Free Account ... glm-4.7:cloud | Available in Free Account ... glm-5.1:cloud | Paid account required ... kimi-k2.6:cloud | Paid account required ... deepseek-v4-pro:cloud | Paid account required"
https://www.reddit.com/r/ollama/comments/1u2ahuj/ollama_free_tier_model_status/ -
"MiniMax-M3 scores 55 on the Artificial Analysis Intelligence Index."
https://artificialanalysis.ai/articles/gemma-4-everything-you-need-to-know
Alternatives
Same category · similar quota
Zhipu AI's BigModel open platform exposes permanently free GLM 'Flash' models (GLM-4.5-Flash, GLM-4.7-Flash, GLM-4.6V-Flash vision) with no credit card required.
