Model by Meta

Llama 4 Maverick

Llama 4 Maverick is Meta's 2025 open-weight multimodal model, and it now suits people who need a known, widely hosted model rather than the newest one.

Checked 2026-11-05

Llama 4 Maverick is Meta's 2025 open-weight multimodal model, and it now suits people who need a known, widely hosted model rather than the newest one. The Hugging Face card lists 17B active parameters out of 400B total across 128 experts, a 1M-token context, and a release date of 5 April 2025. It accepts text and images and returns text and code, with 12 officially supported languages. The licence is the Llama 4 Community License, not an open-source one: it requires the Built with Llama credit and asks companies with over 700 million monthly users to apply to Meta. OpenRouter lists a 524,288-token window, half the card's figure, at $0.35 per million input tokens and $1.00 per million output tokens. The card tested image understanding with up to 5 images. Figures come from the Hugging Face card and the OpenRouter model list. We have not run it ourselves. We would pick it only where its licence and hosting already fit, and compare newer models first.

Mentioned here

  • gpt-oss-120b

    gpt-oss-120b is OpenAI's open-weights model under Apache 2.0, a standard permissive licence, with a context of about 131,072 tokens.

    • Open weights

    OpenAI, 131K context, $0.15 in / 1M

  • Mistral Large 4

    Mistral Large 4 is Mistral's flagship open-weights model, released 2026-10-06, and OpenRouter lists it at $0.7 per million input tokens and $2.1 output with a 1,048,576-token context and text and image input.

    • Open weights

    Mistral AI, 1M context, $0.70 in / 1M

  • Meta Llama API

    Meta's API now leads with Muse Wonder $4/$20 and Muse Spark $1.25/$4.25. Contributor mode trades training rights for a discount. Limits and agent setup.

    • Model / inference API
    • OpenAI compatible

  • OpenRouter

    OpenRouter routes more than 600 models through one API key.

    • Model gateway (router)
    • OpenAI compatible

  • Groq

    GroqCloud serves open models on its own inference hardware, with speed as the draw.

    • Model / inference API
    • OpenAI compatible

Questions

What is the context window of Llama 4 Maverick?

524,288 tokens, with up to 16,384 tokens of output.

How much does Llama 4 Maverick cost?

$0.35 per million input tokens and $1 per million output tokens, checked 5 November 2026.

Can I run Llama 4 Maverick on my own hardware?

Yes, the weights are published under the Open-weights model licence. They are on huggingface.co.

Who makes Llama 4 Maverick?

Meta makes it, and released it on 5 April 2025.

Reviews

Loading reviews…

Write a review

Your rating

Up to 1000 characters. Stars alone are enough.

Similar

  • Muse Spark 1.3

    Muse Spark 1.3 is Meta's closed-weights reasoning model for agentic and coding work, and it suits developers who want a 1 million token context from a vendor outside the usual Anthropic, OpenAI and Google trio.

    • Closed

    Meta, 1M context, $1.25 in / 1M

  • Muse Wonder

    Muse Wonder is a frontier reasoning model from Meta, aimed at complex personal and professional agent tasks.

    • Closed

    Meta, 1M context, $4 in / 1M

  • Muse Spark Open

    Muse Spark Open is Meta's open-weights model in its new Muse line, released 2026-10-15, listed on OpenRouter at $1.25 per million input tokens and $4.25 output, with a 1,048,576-token context and text, image, video, PDF and audio input.

    • Open weights

    Meta, 1M context, $1.25 in / 1M