Ollama

Ollama is a local model runner with an optional cloud. Local runs on your hardware stay unlimited on every plan. Cloud usage is metered; Pro is 50x Free, Max is 5x Pro, with session limits every 5 hours and weekly limits every 7 days. github.com/ollama/ollama had 178,742 stars, license MIT. The old three-paragraph "install Llama3-8B-Chinese-Chat in 30 minutes" body is out.

Compare OpenRouter if you wanted a multi-vendor API router, and LangChain if you only needed an orchestration layer on top of a local endpoint.

Key Features

Local first: Download the app or CLI and run models on your machine. That path does not consume the cloud quota table.
Cloud extras: Paid rows unlock larger hosted models, more concurrency (Free 1 / Pro 3 / Max 10), and extra usage packs on Pro/Max.
Team waitlist: Team is introductory at $25 / seat, 5-seat minimum, shared billing. SSO is listed as coming soon.
Model catalog: The GitHub blurb the day we checked mentioned Kimi K2.6, GLM-5.2, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma. Treat ollama.com as the live list.

Use Cases

People who will keep weights on a laptop and stay on Free $0.
People who will pay Pro $20 for hosted larger models and 3-way concurrency.
Teams comparing Team $25 x 5 vs five Pro seats.

Limitation: Domain rating 165 / authority 160 in frontmatter are leftover catalog metrics, not a new audit. We did not re-verify "40,000+ integrations." Max is paused for new buyers.

Pricing

ollama.com/pricing on 2026-08-17.

Plan	Price	Notes from that page
Free	$0	Local unlimited. Light cloud. 1 concurrent cloud model.
Pro	$20 / month or $200 / year	50x Free cloud usage. 3 concurrent. Extra usage available.
Max	$100 / month	5x Pro. 10 concurrent. New sign-ups paused.
Team	$25 / seat / month	5-seat minimum. Usage included per seat, then shared overage. Waitlist.
Enterprise	Custom	Volume, security, procurement.

If a leftover blog still describes Ollama as only a 30-minute local installer, treat this table as the source.

Getting Started

Download from ollama.com and run one local model before you buy Pro.
Open ollama.com/pricing and read the 5-hour / 7-day cloud caps.
Do not join Max while new sign-ups are paused.
Confirm OpenRouter if you needed many vendors, not one local runner.

Frequently Asked Questions

Is Ollama still free-only?

Local is free. Cloud has Free / Pro / Max / Team rows.

Is Max $100 available?

Existing Max stays.

Same as OpenRouter?

No. Ollama is a local runner plus its own cloud. OpenRouter routes many providers.

Alternatives

OpenRouter: Multi-model API router.
LangChain: Orchestration on top of a local or cloud endpoint.

Tips

Quote Pro $20 / month or $200 / year first.
Do not paste the Llama3-8B-Chinese-Chat 30-minute tutorial as the product.
Confirm Max pause before you promise a $100 seat.

Conclusion

Ollama is a MIT local runner with a public Free / Pro / Max / Team cloud ladder, not a generic 30-minute install essay. Start at ollama.com/pricing, then decide whether OpenRouter already covers the API you need.