Not affiliated with Meta. Independent community review.

releasedv4 / 4 Maverick

Llama Review (2026)

Open-source foundation model for self-hosting

Last updated: 2026-02-08

Key Conclusions

  1. 1

    Fully open-source with Meta's permissive community license.

  2. 2

    Llama 4 Scout/Maverick introduces MoE architecture with massive 10M context.

  3. 3

    Zero API cost for self-hosted deployments; cloud hosting available via partners.

Parameters

Context Window

10M tokens

Max Output

64K tokens

Multimodal

Yes

Languages

200+

Pricing & API

Input Price

$0.00

per 1M tokens

Output Price

$0.00

per 1M tokens

Free TierAvailable

Frequently Asked Questions

What is Llama 4?

Llama 4 is Meta's latest open-source model with mixture-of-experts architecture and up to 10M token context.

Is Llama free?

Yes. Llama models are free to download and use under Meta's community license.

Can I self-host Llama?

Yes. Llama weights are freely available for self-hosting on your own infrastructure.

What hardware do I need to run Llama 4?

Llama 4 Scout (17B active params) can run on a single GPU with 32GB+ VRAM. Maverick (400B+ total) requires multi-GPU setups or cloud instances with 4-8 A100/H100 GPUs.

What is the Llama community license?

Meta's community license allows free use for research and commercial applications with over 700M monthly active users requiring a separate license agreement.

About Llama

Llama is Meta's open-source model family. Llama 4 introduces mixture-of-experts architecture and a 10M token context window.

How We Evaluate

Our reviews are based on publicly available documentation, API specifications, and benchmark results. We evaluate models across five dimensions:

  • Context capacity — maximum input tokens and practical retrieval accuracy at scale.
  • Output quality — coherence, factual accuracy, and instruction following based on published benchmarks.
  • Pricing transparency — clarity of per-token costs, free-tier limits, and hidden fees.
  • Multimodal breadth — native support for text, image, audio, and video inputs/outputs.
  • Ecosystem maturity — SDK quality, documentation depth, and third-party integrations.

Scores and conclusions are updated when providers announce pricing changes or new model versions. This page was last verified on 2026-02-08.

Explore More

Stay Updated on Llama

Get notified when new benchmarks, pricing changes, or major updates drop.