Pricing and plans

There is no Llama subscription. This lesson lays out the three real cost routes: the free Meta AI assistant, hosted APIs billed per token, and self-hosting.

In this lesson you will

Explain why Llama has no subscription tier and what replaces it
Estimate a monthly hosted-API bill from per-token prices
Decide between hosted API and self-hosting based on volume and privacy

If you came here looking for a "Llama Plus" plan, there is none. Meta does not sell a consumer subscription for Llama, and our catalogue has no Meta subscription rows. What you have instead are three routes with very different cost shapes.

Figure 1.Llama has no subscription tier. Pick between free-but-limited, metered, or fixed-cost hardware.

Route 1: Meta AI, free

Meta AI is the assistant built into Meta's apps and available at meta.ai. It costs nothing. For a beginner who wants to ask questions, draft messages or summarise a paragraph, it is a perfectly good place to start. What it does not give you is model choice, an API, a system prompt or any control over settings.

Route 2: hosted API, per token

A hosted API is the middle path and the one most of this course uses. You create an account with a provider, add a small amount of credit, and pay only for the tokens you send and receive. At the time of writing OpenRouter lists Llama 4 Scout at $0.10 per million input tokens and $0.30 per million output tokens in our catalogue.

That is cheap enough that a casual user rarely spends more than pocket change. A thousand short chats a month, at roughly 500 input and 500 output tokens each, is about half a million tokens each way. On Scout's rates that is around $0.20 for the month. Put your own numbers into the cost calculator to check.

Figure 2.A flat subscription costs the same whether you use it or not. Per-token billing tracks your actual usage. Llama only offers the second shape.

Route 3: self-host, fixed cost

Self-hosting flips the cost shape. There is no meter, but you pay for a GPU up front or rent one by the hour, and you pay for power and your own time keeping it running. It is worth it when your volume is very high, when your data cannot leave your network, or when you want to fine-tune. It is not worth it for a few hundred prompts a month.

What each route is worth

Casual user: Meta AI, free. You lose nothing by starting here.
Hobbyist or developer: hosted API. Cheap, flexible, no commitment. Compare it with a $20-class subscription in subscription vs API cost.
Team with volume or privacy needs: self-host, or a hosted API with a business agreement. Get help from the decision engine.

Knowledge check

Which statement about Llama pricing is correct?

Lesson FAQ

Is there a Llama subscription like ChatGPT Plus?

No. Meta AI is free, and everything else is metered per token through a host or paid for as hardware if you self-host.

How much does the Llama API cost per month?

It depends entirely on tokens used. At the time of writing Llama 4 Scout is $0.10 per million input tokens on OpenRouter in our catalogue. Light personal use is typically well under a dollar a month.

When should I self-host Llama?

When your hosted bill is large enough to rival hardware cost, when data cannot leave your network, or when you need to fine-tune. For most beginners a hosted API is cheaper and simpler.