Llama

Name: Llama
Brand: Meta

Meta, United States, dev.meta.ai

Open AI models released by Meta. You can download them and build them into your own service

Open Models

Go to official siteRequest a correction

CurrencyUSD KRW EUR

Personal starting priceFreeNo paid plan

Free tierFree

Korean supportAverage

Price checked 2026-10-19, source: dev.meta.ai. Prices and benefits may change. Check the official site before you pay.

Pros

Download and use for free
Very large ecosystem and many resources

Cons

Running it yourself needs a server and GPU
License terms need checking
No new version since April 2025

These are opinions compiled by Pickx from public sources.

At a glance

Llama is a series of open-weight AI models released by Meta. Anyone can download them, run them on their own server/PC, or fine-tune them and build them into a service. However, no new Llama has appeared since Llama 4 (Scout/Maverick) in April 2025. From 2026 Meta shifted to the Muse series and released Muse Glimmer (30B, Apache 2.0) as a successor open model. llama.com now links to Meta's developer platform (dev.meta.ai). It suits developers/startups/researchers, and companies that do not want to send data outside.

Full model list

Model	Type	Features	Plans that can use it
Llama 4 Maverick (17B active 400B, 128 experts)	Text/image input (MoE)	Top released Llama model 1M token context Strong at image understanding/multilingual/coding Needs a large GPU server to run Released April 2025 Knowledge cutoff August 2024	Free download Cloud API
Llama 4 Scout (17B active 109B, 16 experts)	Text/image input (MoE)	10M token context, among the longest in the industry Strong at analyzing long documents/codebases Can run on 1 GPU (H100) when quantized	Free download Cloud API
Llama 4 Behemoth	Text/multimodal	Huge teacher model announced in 2025 Not released	Not released
Llama 3.3 70B Instruct	Text	128K context Value model with performance close to the 405B class at 70B size Released December 2024	Free download Cloud API (previous model)
Llama 3.2 1B 3B	Text (lightweight)	128K context Ultra-light models that run on devices such as smartphones/laptops	Free download (previous model)
Llama 3.2 11B 90B Vision	Text/image input	Llama's first image understanding models For analyzing documents/charts/photos	Free download Cloud API (previous model)
Llama 3.1 8B 70B 405B	Text	128K context Officially supports 8 languages 405B was the largest released model at the time Widely used for synthetic data generation/distillation Released July 2024	Free download Cloud API (previous model)
Llama 3 8B 70B	Text	8K context First models of the 3rd generation, released April 2024	Free download (previous model)
Llama 2 7B 13B 70B Code Llama 7B to 70B	Text/coding	4K context (Code Llama up to 100K) First generation to allow commercial use Released 2023	Free download (previous model)
Llama Guard 4 (12B) Prompt Guard 2 (22M/86M)	Safety classification	Checks inputs/outputs for harmfulness Auxiliary models for detecting prompt injection/jailbreaks Released April 2025	Free download
Muse Glimmer (about 30B, for reference)	Text/image input (Dense)	Meta open model released August 2026 as a Llama successor 131K context 100 or more languages Runs on a 24 to 32GB GPU with 4-bit quantization Apache 2.0, so no limit on number of users	Free download (Hugging Face)
Muse Spark Open (816B total 42B active, for reference)	Multimodal (MoE)	Meta open-weight model released 2026-10-15 1,000,000 token context For coding and long agent tasks Muse Model License (commercial use has conditions) Can run locally on 2 Mac Studio (M5 Ultra)	Free download (Hugging Face SageMaker Google Cloud Foundry) Cloud API (Together Groq Fireworks and others)

License summary: Llama 2/3/4 can be used commercially for free under the "Llama Community License" of each version. However, companies with more than 700,000,000 monthly active users at the time of release need separate permission from Meta, and when distributing you must show "Built with Llama", put "Llama" at the start of derived model names, and follow the acceptable use policy (AUP). Because models such as the Llama 3.2 multimodal models had restrictions on use in the EU under the usage policy, check each version's license for details. Muse Glimmer is Apache 2.0.

Plans

Plan	Price	What you can do	Usage limits	Features
Free download (Hugging Face/official repository)	$0 (you pay for hardware/electricity)	Download all released Llama models Fine-tune Build into commercial services (within license terms)	No limit on number of uses	Can also run on a PC with Ollama/LM Studio/llama.cpp (small models)
Cloud API (AWS Bedrock, Azure, Google Cloud, Groq, Together and others)	Usage-based billing (varies by provider, needs checking)	Call Llama 4/3.x through an API with no server	Varies by provider (not disclosed officially)	Price/speed/supported models differ by provider
Separate license for large companies	Negotiated individually with Meta	Permission to use for companies with more than 700,000,000 monthly active users	Negotiated individually with Meta	For giant platform companies

There is no paid personal subscription plan. Meta's own API (Meta Model API) currently focuses mainly on the Muse series.

Which plan to choose

Free download (small models: Llama 3.2 1B/3B, 3.1 8B): Recommended for students/developers who want to run AI offline on their own laptop.
Free download (Llama 4 Scout/3.3 70B): Suits companies that must not send customer data outside, or teams with their own GPU servers.
Cloud API: Suits startups/developers who want to add a cheap open model to a service without managing servers.
If you are starting fresh: As of 2026 Llama updates have stopped, so also compare Muse Glimmer, which has a freer license, and other latest open models.

Notes

Information checked: 2026-10-12
Sources: Hugging Face meta-llama organization (official model list), Llama 4 Scout model card/license, Meta developer platform (formerly llama.com), Muse Glimmer model card, Muse Glimmer release coverage (Implicator)

Good for

6 views (2 today)

Prices converted at 1 USD = ₩1,329.49 (2026-10-19). Original price shown in brackets. Taxes as listed by each provider.

Comments

No comments yet. Be the first.

Your comments and nickname are public. If you see abuse or advertising, report it here and we will review and remove it.