Meta: Llama models
Llama is Meta's family of open models — they can run on your own hardware or through providers. What they are good at, limitations and how to connect them in SKILLOK.
In this article
Who they are#
Meta releases the Llama family — models with open weights. They can be downloaded and run on your own hardware or used through cloud providers.
Models#
- Llama of different sizes — from small ones that run on a laptop to large ones for servers.
- Thousands of fine-tuned community models are built on Llama.
The exact list and versions change often — current models are visible in the list when you connect a key in SKILLOK, and in the model catalog.
What they are good at#
- Open weights mean full control: data does not go to a third party if the model runs on your side.
- A large community and many fine-tuned versions for specific tasks.
Access from Russia#
The weights can be downloaded and run anywhere; cloud providers of Llama are available through intermediaries such as OpenRouter.
Models in the catalog#
Prices, context window and comparison with similar models:
- Llama 4 Maverick — $0.188 / $0.652 per million tokens
- Llama 4 Scout — $0.1 / $0.3 per million tokens
- Llama 3.3 70B Instruct — $0.1 / $0.32 per million tokens
How to connect in SKILLOK#
Llama models work in SKILLOK through OpenRouter — a meta-llama/… model. Connecting your own server with a model is planned.
In detail — Connecting in SKILLOK.
Prices#
Prices per million input and output tokens are on the official page. Through OpenRouter the prices are the same or close; how much you spent is shown in Settings → AI.
Was this helpful?