Is Llama 4 Maverick open source?

Meta distributes Llama 4 Maverick as open weights, which means you can download and run it on your own infrastructure as well as buy it through hosted inference providers. That is the main practical difference between it and the closed frontier models it competes with.

Can Llama 4 Maverick read images?

Yes. It accepts both text and image input, so screenshots, charts, documents and product photos can be passed in alongside a text prompt in the same request.

How much can Llama 4 Maverick handle in one request?

The context window runs to over a million tokens, enough to hold a full documentation set or a long batch of transcripts at once. The maximum output is much smaller, so it is built to read a lot and write concisely rather than produce long documents.

When should I choose a different model?

Pick something else when you need a long single generated output, such as a full report or a large code file, or when the task depends on the strongest available reasoning. Maverick is the value choice for high-volume reading and answering, not the ceiling on capability.

Meta model°

Llama 4 Maverick

Llama 4 Maverick is Meta's open-weight multimodal model, released in April 2025, which reads both text and images and runs across a context window of roughly a million tokens. It is the mid-tier instruct model in the Llama 4 family, positioned for teams that want to self-host or buy inference cheaply rather than commit to a closed API.

MetaVerified 24/09/2026Released 05/04/2025

01 / decision

Decision snapshot

Each figure sits next to the middle of the field, so it can be read as dear or cheap, wide or narrow, rather than floating on its own. Compared against the 126 models we track, not against an absolute standard.

Capability

Not measured

No published benchmark scores yet

Input / 1M tokens

$0.19

field median $0.30

Output / 1M tokens

$0.65

field median $1.25

Context

1,049K

field median 524K

02 / overview

What Llama 4 Maverick is, and when to reach for it

Four questions, answered separately, because somebody arrives at one of them rather than at the top of the page.

What it is

The job it was built for, and the job it is not for.

Llama 4 Maverick is an instruction-tuned multimodal language model from Meta, built on a mixture-of-experts design with 128 experts and 17 billion active parameters per forward pass. It is built for general assistant work at volume: answering questions, summarising long documents, reading images alongside text, and powering chat and retrieval products. It is not a specialist reasoning or coding model, and its short maximum output means it is a poor fit for jobs that need long generated artefacts in one pass.

When it arrived, and when to use it

Where it sits in its line, and when a sibling is the better pick.

Maverick first appeared in April 2025 as part of the Llama 4 release. It sits above the smaller Llama 4 variants in capacity and below the largest, which makes it the sensible default when you want multimodal input and a very long context without paying frontier prices. It is the wrong pick when you need a long single response, or when a task genuinely demands the strongest available reasoning.

How you reach it

The API, the apps it powers, and what its limits let you do.

Maverick is reached through Meta's own distribution and through the many hosting providers that serve Llama weights, so it is usually a standard chat completions call with an image field. The context window is large enough to hold an entire documentation set, a long support history or a batch of transcripts in a single request, and image input means screenshots, charts and product photos can go in alongside the text. Output is capped well below the input ceiling, so it reads far more than it writes.

Why it matters

What changes because this exists, or why it does not.

The combination of open weights, multimodal input and a million-token window at low cost changes what is affordable to run at scale. Work that previously meant chunking and re-ranking, such as reading a full knowledge base per query, can now be done by stuffing the context, and teams with data residency or self-hosting requirements get a multimodal option they can run themselves.

Superseded by Llama 4 Scout. See the whole line.

03 / evidence

How much of this is verified

Split by category, so a strong number never hides a thin evidence base. Verified means we read it on the benchmark's own published results; a provider's claim about its own model is shown and labelled rather than dropped.

No published benchmark scores for this model yet.

We publish a score only where we can link the result to where it was published. Until a benchmark result for this model exists in a source we read, this section stays empty rather than being filled with a provider's marketing figure.

How we decide what counts as evidence

04 / ledger

Benchmark ledger

Every published row, grouped by category, each compared with the best published score on the same benchmark. 'Is 64% good' is a question nobody can answer; '26 points behind the leader' is one anybody can.

Nothing in the ledger yet.

Each row here carries a score, the benchmark it came from, what the leading model scored on the same test, and a link to the published result. Rows appear as results are published and read.

How we decide what counts as evidence

05 / capability

Capability shape

Where this model is strong, and against how many peers. Ranks are against models with evidence in that category, not against everything we track: ranking against models nobody tested would rank who published, not who is better.

No category scores to shape yet.

A category score is the weighted mean of the benchmarks published for it. With no published rows there is nothing to average, and an empty chart drawn at zero would say something false.

How we decide what counts as evidence

06 / cost

What it costs

List API rates as last read from the provider, with the source on every row, plus every change we have recorded since we started tracking it.

Llama 4 Maverick API pricingSurge45°
Charge	Price	Unit	Read on	Source
Input	$0.19	per 1M tokens	2026-09-24	Check
Output	$0.65	per 1M tokens	2026-09-24	Check

Maverick is among the cheapest capable multimodal models to run, with input costing a fraction of what closed frontier models charge and output only modestly more than input. For high-volume work where the model reads a great deal and writes little, the economics are hard to beat, and self-hosting the weights removes the per-token cost entirely if you have the infrastructure.

What a month costsSurge45°
Workload	Input / month	Output / month	Cost
A small product team	20M tokens	5M tokens	$7.01
A busy support assistant	200M tokens	40M tokens	$63.60
A document pipeline	1000M tokens	100M tokens	$252.75

List API rates, no caching and no batch discount, which both providers offer and which change the answer a great deal. Treat these as the ceiling, not the bill.

07 / specs

Specifications

As listed by the provider's own catalogue and re-read every few hours. Anything absent is absent there too.

SpecificationSurge45°
Context window	1,048,576 tokens
Maximum output	16,384 tokens
Modalities	text, image
Released	05/04/2025
Status	Current
Catalogue identifier	meta-llama/llama-4-maverick

08 / lineage

Lineage

What this model replaced, what replaced it, and what else its provider has in the field.

Also from Meta

09 / line

The line

Every model in this family in release order, so a page from eight months ago says in one glance that two newer ones exist.

Came before

The first we track in this line.

Came after

Llama 4 Scout →

01Llama 4 Maverick05/04/2025
02Llama 4 Scout05/04/2025

Ordered by release date and worked out from the naming, so a new member slots in as soon as its page exists. A retired model keeps its page and its place in the line.

10 / notes

Our notes

What this model changes for a brand trying to be cited in AI answers, and every change we have logged since it launched.

What it changes for you

When an open-weight model this cheap gets embedded in support bots, internal search and third-party assistants, your brand is being summarised by systems you will never see in a logs dashboard. Maverick's enormous context window means those systems can read your entire documentation site in one request, so the depth and structure of your public content matters more than clever snippets. Make sure the pages that explain what you do, who you serve and how you compare are crawlable, unambiguous and complete, because that is the whole of what the model has to go on.

Where buyers meet this model

Buyers meet Maverick inside Meta's own assistant surfaces and, more often, without knowing it, as the model behind third-party chat products, support bots and search tools that chose an open-weight option for cost reasons. On the API side it is available through most of the major inference hosts, which is where engineering teams encounter it when comparing per-token costs.

Change log

Nothing published here yet. Changes appear within hours of a provider announcing them.

11 / questions

Questions

The things people ask about this model, answered from what is on this page rather than from anywhere else.

Is Llama 4 Maverick open source?: Meta distributes Llama 4 Maverick as open weights, which means you can download and run it on your own infrastructure as well as buy it through hosted inference providers. That is the main practical difference between it and the closed frontier models it competes with.
Can Llama 4 Maverick read images?: Yes. It accepts both text and image input, so screenshots, charts, documents and product photos can be passed in alongside a text prompt in the same request.
How much can Llama 4 Maverick handle in one request?: The context window runs to over a million tokens, enough to hold a full documentation set or a long batch of transcripts at once. The maximum output is much smaller, so it is built to read a lot and write concisely rather than produce long documents.
When should I choose a different model?: Pick something else when you need a long single generated output, such as a full report or a large code file, or when the task depends on the strongest available reasoning. Maverick is the value choice for high-volume reading and answering, not the ceiling on capability.

Surge45°

Is Llama 4 Maverick recommending you?

Models change what gets cited. We measure whether AI answers name your brand or your competitors across every assistant, and show you what to change.

Free AI visibility audit Book a call