Skip to content
Meta

Llama 4 Maverick

meta/llama-4-maverick
Open weightsLanguage

The open flagship of the Llama 4 generation, multimodal, million-token, permissively licensed.

Model overview

Context window

1M

tokens

Input price

—

quoted on request

Output price

—

quoted on request

Weekly volume

—

tokens / week

Open weights

Yes

self-hostable

Capabilities

5

of 11 tags

Description

Llama 4 Maverick is a language model from Meta. Webparam does not stock this model yet and can source it on request — tell us what you need it for and we will come back with availability and a price. Meta’s stated focus is open weights at the frontier (Llama). Its mean it can also be self-hosted under its licence.

Capabilities

Tool calling
Calls functions you define and returns their arguments as structured data, the basis of agents.
JSON mode
Constrains the response to valid JSON matching a schema you supply.
Streaming
Emits tokens as they are generated, so answers appear progressively rather than all at once.
Open weights
The parameters are published, so the model can be inspected, fine-tuned and self-hosted under its licence.
Vision input
Accepts images alongside text in the prompt.

Strengths & weaknesses

Strengths

  • Open weights with vision
  • 1M context
  • Huge fine-tuning ecosystem

Weaknesses

  • Peak quality trails closed flagships
  • Tool calling less battle-tested than GPT/Claude

Pricing

Pricing for Llama 4 Maverick, per 1M tokens
RatePriceUnit
Input— quoted on request—
Output— quoted on request—

This model is not provisioned yet; the rate is quoted on request.

Context window

1Mtokens

covers prompt and response together, so a long input leaves less room for the answer.

1M tokens against a peer range of 32K to 10M, median 256K. Compared across 34 language models. Logarithmic scale.

Supported features

Feature support for Llama 4 Maverick
FeatureSupport
Tool callingSupported
JSON modeSupported
StreamingSupported
Vision inputSupported
Audio inputNot supported
Long contextNot supported
Open weightsSupported
ReasoningNot supported
Fine-tunableNot supported
BatchNot supported
CachingNot supported

Example use cases

  • Self-host pipelines

  • Fine-tuned products

  • Open-stack RAG

Related models

Comparison

Compare against anything →
Llama 4 Maverick compared with Qwen3 235B A22B Instruct 2507 and Llama 4 Scout
AttributeLlama 4 MaverickQwen3 235B A22B Instruct 2507Llama 4 Scout
Context window1M256K10M
Input priceR0.00R0.00R0.00
Output price— not charged— not charged— not charged
Tool callingSupportedSupportedSupported
JSON modeSupportedSupportedNot supported
Vision inputSupportedNot supportedNot supported
Open weightsSupportedSupportedSupported
Prompt cachingNot supportedNot supportedNot supported

Best value in each row is highlighted. Blank cells are facts the supplier does not publish, not zeroes.

Provider information

View Meta →
Meta2 models

Meta · Menlo Park, United States · est. 2004 (FAIR/Llama programme from 2023)

Meta made open weights a frontier strategy. Llama downloads number in the hundreds of millions, and the Llama 4 generation pushed open models to 10M-token context, the largest window on the market.

Recent releases

    Documentation

    Frequently asked questions

    Input is billed at R0.00 per 1M tokens, and there is no separate output charge. Every figure on this page is an illustrative placeholder.