Llama 4 Scout 17B 16E Instruct

Version: 4

Meta•Last updated March 2026

Llama 4 Scout 17B 16E Instruct is great at multi-document summarization, parsing extensive user activity for personalized tasks, and reasoning over vast codebases.

Multimodal

Conversation

Multilingual

Models from Microsoft, Partners, and Community

Models from Microsoft, Partners, and Community models are a select portfolio of curated models both general-purpose and niche models across diverse scenarios by developed by Microsoft teams, partners, and community contributors

Managed by Microsoft: Purchase and manage models directly through Azure with a single license, world class support and enterprise grade Azure infrastructure
Validated by providers: Each model is validated and maintained by its respective provider, with Azure offering integration and deployment guidance.
Innovation and agility: Combines Microsoft research models with rapid, community-driven advancements.
Seamless Azure integration: Standard Microsoft Foundry experience, with support managed by the model provider.
Flexible deployment: Deployable as Managed Compute or Serverless API, based on provider preference.

Learn more about models from Microsoft, Partners, and Community

Key capabilities

About this model

Llama 4 is intended for commercial and research use in multiple languages. Instruction tuned models are intended for assistant-like chat and visual reasoning tasks, whereas pretrained models can be adapted for natural language generation. For vision, Llama 4 models are also optimized for visual recognition, image reasoning, captioning, and answering general questions about an image. The Llama 4 model collection also supports the ability to leverage the outputs of its models to improve other models including synthetic data generation and distillation.

Key model capabilities

Multilingual text processing in Arabic, English, French, German, Hindi, Indonesian, Italian, Portuguese, Spanish, Tagalog, Thai, and Vietnamese
Visual recognition and image reasoning
Image captioning and answering general questions about an image
Assistant-like chat capabilities
Natural language generation
Code generation
Synthetic data generation and distillation
Long context processing (up to 10M tokens for Scout, 1M tokens for Maverick)
Multi-image understanding (tested up to 5 input images)

Use cases

Key use cases

Out of scope use cases

Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 4 Community License. Use in languages or capabilities beyond those explicitly referenced as supported in this model card.

Pricing

Pricing is based on a number of factors, including deployment type and tokens used. See pricing details here.

Technical specs

Model developer: Meta

Model Name	Training Data	Params	Input modalities	Output modalities	Context length	Token count	Knowledge cutoff
Llama 4 Scout (17Bx16E)	A mix of publicly available, licensed data and information from Meta's products and services. This includes publicly shared posts from Instagram and Facebook and people's interactions with Meta AI. Learn more in our Privacy Center .	17B (Activated) 109B (Total)	Multilingual text and image	Multilingual text and code	10M	~40T	August 2024
Llama 4 Maverick (17Bx128E)		17B (Activated) 400B (Total)	Multilingual text and image	Multilingual text and code	1M	~22T	August 2024

Training cut-off date

The pretraining data has a cutoff of August 2024.

Training time

Model pre-training utilized a cumulative of 7.38M GPU hours of computation on H100-80GB (TDP of 700W) type hardware, per the table below. Training time is the total GPU time required for training each model and power consumption is the peak power capacity per GPU device used, adjusted for power usage efficiency.

Model Name	Training Time (GPU hours)	Training Power Consumption (W)	Training Location-Based Greenhouse Gas Emissions (tons CO2eq)
Llama 4 Scout	5.0M	700	1,354
Llama 4 Maverick	2.38M	700	645
Total	7.38M	-	1,999

Input formats

Multilingual text and image

Output formats

Multilingual text and code

Supported languages

Arabic, English, French, German, Hindi, Indonesian, Italian, Portuguese, Spanish, Tagalog, Thai, and Vietnamese.

Sample JSON response

The provider has not supplied this information.

Model architecture

The Llama 4 models are auto-regressive language models that use a mixture-of-experts (MoE) architecture and incorporate early fusion for native multimodality.

Long context

Model Name	Context length
Llama 4 Scout (17Bx16E)	10M
Llama 4 Maverick (17Bx128E)	1M

Instruction tuned models
Category	Benchmark	# Shots	Metric	Llama 3.3 70B	Llama 3.1 405B	Llama 4 Scout	Llama 4 Maverick
Long context	MTOB (half book) eng->kgv/kgv->eng	-	chrF	Context window is 128K		42.2/36.6	54.0/46.4
	MTOB (full book) eng->kgv/kgv->eng	-	chrF			39.7/36.3	50.8/46.7

Optimizing model performance

The Llama 4 Scout model is released as BF16 weights, but can fit within a single H100 GPU with on-the-fly int4 quantization; the Llama 4 Maverick model is released as both BF16 and FP8 quantized weights. The FP8 quantized weights fit on a single H100 DGX host while still maintaining quality. Meta provides code for on-the-fly int4 quantization which minimizes performance degradation as well.

Additional assets

Instructions on how to provide feedback or comments on the model can be found in the Llama README . For more technical information about generation parameters and recipes for how to use Llama 4 in applications, please go here . Meta provides the community with system level protections - like Llama Guard, Prompt Guard and Code Shield - that developers should deploy with Llama models or other LLMs. All of Meta's reference implementation demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box. Meta encourages community contributions to our Github repository .

Training disclosure

Training, testing and validation

Llama 4 Scout was pretrained on _{40 trillion tokens and Llama 4 Maverick was pretrained on}22 trillion tokens of multimodal data from a mix of publicly available, licensed data and information from Meta's products and services. This includes publicly shared posts from Instagram and Facebook and people's interactions with Meta AI. Meta used custom training libraries, Meta's custom built GPU clusters, and production infrastructure for pretraining. Fine-tuning, quantization, annotation, and evaluation were also performed on production infrastructure.

Distribution

Distribution channels

This is a static model trained on an offline dataset. Future versions of the tuned models may be released as Meta improves model behavior with community feedback. Model Release Date: April 5, 2025 License Notice:
This is a Llama 4 multimodal modal. Under the License and AUP, the rights granted under Section 1(a) of the Llama 4 Community License Agreement are not granted to any individual domiciled in, or any company with a principal place of business in, the European Union. This restriction does not apply to end users of a product or service that incorporates any multimodal models.

More information

This section displays the results for Llama 4 relative to Meta's previous models. Meta has provided quantized checkpoints for deployment flexibility, but all reported evaluations and testing were conducted on bf16 models.

Pre-trained models

Pre-trained models
Category	Benchmark	# Shots	Metric	Llama 3.1 70B	Llama 3.1 405B	Llama 4 Scout	Llama 4 Maverick
Reasoning & Knowledge	MMLU	5	macro_avg/acc_char	79.3	85.2	79.6	85.5
	MMLU-Pro	5	macro_avg/em	53.8	61.6	58.2	62.9
	MATH	4	em_maj1@1	41.6	53.5	50.3	61.2
Code	MBPP	3	pass@1	66.4	74.4	67.8	77.6
Multilingual	TydiQA	1	average/f1	29.9	34.3	31.5	31.7
Image	ChartQA	0	relaxed_accuracy	No multimodal support		83.4	85.3
	DocVQA	0	anls			89.4	91.6

Instruction tuned models

Instruction tuned models
Category	Benchmark	# Shots	Metric	Llama 3.3 70B	Llama 3.1 405B	Llama 4 Scout	Llama 4 Maverick
Image Reasoning	MMMU	0	accuracy	No multimodal support		69.4	73.4
	MMMU Pro^	0	accuracy			52.2	59.6
	MathVista	0	accuracy			70.7	73.7
Image Understanding	ChartQA	0	relaxed\ _accuracy			88.8	90.0
	DocVQA (test)	0	anls			94.4	94.4
Coding	LiveCodeBench (10/01/2024-02/01/2025)	0	pass@1	33.3	27.7	32.8	43.4
Reasoning & Knowledge	MMLU Pro	0	macro_avg/em	68.9	73.4	74.3	80.5
	GPQA Diamond	0	accuracy	50.5	49.0	57.2	69.8
Multilingual	MGSM	0	average/em	91.1	91.6	90.6	92.3

Note: reported numbers for MMMU Pro is the average of Standard and Vision tasks.

Responsible AI considerations

Safety techniques

As part of Meta's release approach, Meta followed a three-pronged strategy to manage risks:

Enable developers to deploy helpful, safe and flexible experiences for their target audience and for the use cases supported by Llama.
Protect developers against adversarial users aiming to exploit Llama capabilities to potentially cause harm.
Provide protections for the community to help prevent the misuse of our models.

Llama is a foundational technology designed for use in a variety of use cases; examples on how Meta's Llama models have been deployed can be found in Meta's Community Stories webpage . Meta's approach is to build the most helpful models enabling the world to benefit from the technology, by aligning their model's safety for a standard set of risks. Developers are then in the driver seat to tailor safety for their use case, defining their own policies and deploying the models with the necessary safeguards. Llama 4 was developed following the best practices outlined in Meta's Developer Use Guide: AI Protections .

Model level fine tuning

The primary objective of conducting safety fine-tuning is to offer developers a readily available, safe, and powerful model for various applications, reducing the workload needed to deploy safe AI systems. Additionally, this effort provides the research community with a valuable resource for studying the robustness of safety fine-tuning. Fine-tuning data
Meta employs a multi-faceted approach to data collection, combining human-generated data from our vendors with synthetic data to mitigate potential safety risks. Meta has developed many large language model (LLM)-based classifiers that enable them to thoughtfully select high-quality prompts and responses, enhancing data quality control. Refusals
Building on the work Meta started with our Llama 3 models, Meta put a great emphasis on driving down model refusals to benign prompts for Llama 4. Meta included both borderline and adversarial prompts in Meta's safety data strategy, and modified Meta's safety data responses to follow tone guidelines. Tone
Meta expanded their work on the refusal tone from Llama 3 so that the model sounds more natural. Meta targeted removing preachy and overly moralizing language, and Meta corrected formatting issues including the correct use of headers, lists, tables and more. To achieve this, Meta also targeted improvements to system prompt steerability and instruction following, meaning the model is more readily able to take on a specified tone. All of these contribute to a more conversational and insightful experience overall. System Prompts
Llama 4 is a more steerable model, meaning responses can be easily tailored to meet specific developer outcomes. Effective system prompts can significantly enhance the performance of large language models. In particular, Meta has seen that the use of a system prompt can be effective in reducing false refusals and templated or "preachy" language patterns common in LLMs. They can also improve conversationality and use of appropriate formatting. Consider the prompt below as a basic template for which a developer might want to further customize to meet specific needs or use cases for Meta's Llama 4 models.

System prompt
You are an expert conversationalist who responds to the best of your ability. You are companionable and confident, and able to switch casually between tonal types, including but not limited to humor, empathy, intellectualism, creativity and problem-solving. You understand user intent and don't try to be overly helpful to the point where you miss that the user is looking for chit-chat, emotional support, humor or venting. Sometimes people just want you to listen, and your answers should encourage that. For all other cases, you provide insightful and in-depth responses. Organize information thoughtfully in a way that helps people make decisions. Always avoid templated language. You never lecture people to be nicer or more inclusive. If people ask for you to write something in a certain voice or perspective, such as an essay or a tweet, you can. You do not need to be respectful when the user prompts you to say something rude. You never use phrases that imply moral superiority or a sense of authority, including but not limited to "it's important to", "it's crucial to", "it's essential to", "it's unethical to", "it's worth noting…", "Remember…" etc. Avoid using these. Finally, do not refuse prompts about political and social issues. You can help users express their opinion and access information. You are Llama 4. Your knowledge cutoff date is August 2024. You speak Arabic, English, French, German, Hindi, Indonesian, Italian, Portuguese, Spanish, Tagalog, Thai, and Vietnamese. Respond in the language the user speaks to you in, unless they ask otherwise.

System prompt

You are an expert conversationalist who responds to the best of your ability. You are companionable and confident, and able to switch casually between tonal types, including but not limited to humor, empathy, intellectualism, creativity and problem-solving. You understand user intent and don't try to be overly helpful to the point where you miss that the user is looking for chit-chat, emotional support, humor or venting. Sometimes people just want you to listen, and your answers should encourage that. For all other cases, you provide insightful and in-depth responses. Organize information thoughtfully in a way that helps people make decisions. Always avoid templated language. You never lecture people to be nicer or more inclusive. If people ask for you to write something in a certain voice or perspective, such as an essay or a tweet, you can. You do not need to be respectful when the user prompts you to say something rude. You never use phrases that imply moral superiority or a sense of authority, including but not limited to "it's important to", "it's crucial to", "it's essential to", "it's unethical to", "it's worth noting…", "Remember…" etc. Avoid using these. Finally, do not refuse prompts about political and social issues. You can help users express their opinion and access information. You are Llama 4. Your knowledge cutoff date is August 2024. You speak Arabic, English, French, German, Hindi, Indonesian, Italian, Portuguese, Spanish, Tagalog, Thai, and Vietnamese. Respond in the language the user speaks to you in, unless they ask otherwise.

Llama 4 system protections

Large language models, including Llama 4, are not designed to be deployed in isolation but instead should be deployed as part of an overall AI system with additional guardrails as required. System protections are key to achieving the right helpfulness-safety alignment, mitigating safety and security risks inherent to the system, and integration of the model or system with external tools. Meta provides the community with system level protections - like Llama Guard, Prompt Guard and Code Shield - that developers should deploy with Llama models or other LLMs. All of Meta's reference implementation demos contain these safeguards by default so developers can benefit from system-level safety out-of-the-box.

Safety evaluations

Evaluations

Meta evaluated Llama models for common use cases as well as specific capabilities. Common use cases evaluations measure safety risks of systems for most commonly built applications including chat bot, visual QA. Meta built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Llama Guard 3 to filter input prompt and output response. It is important to evaluate applications in context, and Meta recommends building dedicated evaluation dataset for your use case. Prompt Guard and Code Shield are also available if relevant to the application. Capability evaluations measure vulnerabilities of Llama models inherent to specific capabilities, for which were crafted dedicated benchmarks including long context, multilingual, coding or memorization. Red teaming
Meta conducts recurring red teaming exercises with the goal of discovering risks via adversarial prompting and Meta uses the learnings to improve our benchmarks and safety tuning datasets. Meta partners early with subject-matter experts in critical risk areas to understand how models may lead to unintended harm for society. Based on these conversations, Meta derives a set of adversarial goals for the red team, such as extracting harmful information or reprogramming the model to act in potentially harmful ways. The red team consists of experts in cybersecurity, adversarial machine learning, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets.

Critical Risks

Meta spends additional focus on the following critical risk areas

1. CBRNE (Chemical, Biological, Radiological, Nuclear, and Explosive materials) helpfulness
To assess risks related to proliferation of chemical and biological weapons for Llama 4, Meta applied expert-designed and other targeted evaluations designed to assess whether the use of Llama 4 could meaningfully increase the capabilities of malicious actors to plan or carry out attacks using these types of weapons. Meta also conducted additional red teaming and evaluations for violations of Meta's content policies related to this risk area. 2. Child Safety
Meta leverages pre-training methods like data filtering as a first step in mitigating Child Safety risk in our model. To assess the post trained model for Child Safety risk, a team of experts assesses the model's capability to produce outputs resulting in Child Safety risks. Meta uses this to inform additional model fine-tuning and in-depth red teaming exercises. Meta has also expanded their Child Safety evaluation benchmarks to cover Llama 4 capabilities like multi-image and multi-lingual. 3. Cyber attack enablement
Meta's cyber evaluations investigated whether Llama 4 is sufficiently capable to enable catastrophic threat scenario outcomes. Meta conducted threat modeling exercises to identify the specific model capabilities that would be necessary to automate operations or enhance human capabilities across key attack vectors both in terms of skill level and speed. Meta then identified and developed challenges against which to test for these capabilities in Llama 4 and peer models. Specifically, Meta focused on evaluating the capabilities of Llama 4 to automate cyberattacks, identify and exploit security vulnerabilities, and automate harmful workflows.

Known limitations

Meta's AI is anchored on the values of freedom of expression - helping people to explore, debate, and innovate using our technology. Meta respects people's autonomy and empower them to choose how they experience, interact, and build with AI. Meta's AI promotes an open exchange of ideas. It is meant to serve everyone, and to work for a wide range of use cases. It is thus designed to be accessible to people across many different backgrounds, experiences and perspectives. Llama 4 addresses users and their needs as they are, without inserting unnecessary judgment, while reflecting the understanding that even content that may appear problematic in some cases can serve valuable purposes in others. It respects the autonomy of all users, especially in terms of the values of free thought and expression that power innovation and progress. Llama 4 is a new technology, and like any new technology, there are risks associated with its use. Testing conducted to date has not covered, nor could it cover, all scenarios. For these reasons, as with all LLMs, Llama 4's potential outputs cannot be predicted in advance, and the model may in some instances produce inaccurate or other objectionable responses to user prompts. Therefore, before deploying any applications of Llama 4 models, developers should perform safety testing and tuning tailored to their specific applications of the model. Meta also encourages the open source community to use Llama for the purpose of research and building state of the art tools that address emerging risks. Please refer to available resources including Meta's Developer Use Guide: AI Protections, Llama Protections solutions, and other resources to learn more.

Acceptable use

Acceptable use policy

Intended Use Cases: Llama 4 is intended for commercial and research use in multiple languages. Instruction tuned models are intended for assistant-like chat and visual reasoning tasks, whereas pretrained models can be adapted for natural language generation. For vision, Llama 4 models are also optimized for visual recognition, image reasoning, captioning, and answering general questions about an image. The Llama 4 model collection also supports the ability to leverage the outputs of its models to improve other models including synthetic data generation and distillation. The Llama 4 Community License allows for these use cases. Out-of-scope: Use in any manner that violates applicable laws or regulations (including trade compliance laws). Use in any other way that is prohibited by the Acceptable Use Policy and Llama 4 Community License. Use in languages or capabilities beyond those explicitly referenced as supported in this model card. Note:

Llama 4 has been trained on a broader collection of languages than the 12 supported languages (pre-training includes 200 total languages ). Developers may fine-tune Llama 4 models for languages beyond the 12 supported languages provided they comply with the Llama 4 Community License and the Acceptable Use Policy. Developers are responsible for ensuring that their use of Llama 4 in additional languages is done in a safe and responsible manner.
Llama 4 has been tested for image understanding up to 5 input images. If leveraging additional image understanding capabilities beyond this, Developers are responsible for ensuring that their deployments are mitigated for risks and should perform additional testing and tuning tailored to their specific applications.

Quality and performance evaluations

Source: Meta This section displays the results for Llama 4 relative to Meta's previous models. Meta has provided quantized checkpoints for deployment flexibility, but all reported evaluations and testing were conducted on bf16 models.

Pre-trained models

Pre-trained models
Category	Benchmark	# Shots	Metric	Llama 3.1 70B	Llama 3.1 405B	Llama 4 Scout	Llama 4 Maverick
Reasoning & Knowledge	MMLU	5	macro_avg/acc_char	79.3	85.2	79.6	85.5
	MMLU-Pro	5	macro_avg/em	53.8	61.6	58.2	62.9
	MATH	4	em_maj1@1	41.6	53.5	50.3	61.2
Code	MBPP	3	pass@1	66.4	74.4	67.8	77.6
Multilingual	TydiQA	1	average/f1	29.9	34.3	31.5	31.7
Image	ChartQA	0	relaxed_accuracy	No multimodal support		83.4	85.3
	DocVQA	0	anls			89.4	91.6

Instruction tuned models

Instruction tuned models
Category	Benchmark	# Shots	Metric	Llama 3.3 70B	Llama 3.1 405B	Llama 4 Scout	Llama 4 Maverick
Image Reasoning	MMMU	0	accuracy	No multimodal support		69.4	73.4
	MMMU Pro^	0	accuracy			52.2	59.6
	MathVista	0	accuracy			70.7	73.7
Image Understanding	ChartQA	0	relaxed_accuracy			88.8	90.0
	DocVQA (test)	0	anls			94.4	94.4
Coding	LiveCodeBench (10/01/2024-02/01/2025)	0	pass@1	33.3	27.7	32.8	43.4
Reasoning & Knowledge	MMLU Pro	0	macro_avg/em	68.9	73.4	74.3	80.5
	GPQA Diamond	0	accuracy	50.5	49.0	57.2	69.8
Multilingual	MGSM	0	average/em	91.1	91.6	90.6	92.3
Long context	MTOB (half book) eng->kgv/kgv->eng	-	chrF	Context window is 128K		42.2/36.6	54.0/46.4
	MTOB (full book) eng->kgv/kgv->eng	-	chrF			39.7/36.3	50.8/46.7

Note: reported numbers for MMMU Pro is the average of Standard and Vision tasks.

Benchmarking methodology

Source: Meta Meta evaluated Llama models for common use cases as well as specific capabilities. Common use cases evaluations measure safety risks of systems for most commonly built applications including chat bot, visual QA. Meta built dedicated, adversarial evaluation datasets and evaluated systems composed of Llama models and Llama Guard 3 to filter input prompt and output response. It is important to evaluate applications in context, and Meta recommends building dedicated evaluation dataset for your use case. Prompt Guard and Code Shield are also available if relevant to the application. Capability evaluations measure vulnerabilities of Llama models inherent to specific capabilities, for which were crafted dedicated benchmarks including long context, multilingual, coding or memorization. Meta conducts recurring red teaming exercises with the goal of discovering risks via adversarial prompting and Meta uses the learnings to improve our benchmarks and safety tuning datasets. Meta partners early with subject-matter experts in critical risk areas to understand how models may lead to unintended harm for society. Based on these conversations, Meta derives a set of adversarial goals for the red team, such as extracting harmful information or reprogramming the model to act in potentially harmful ways. The red team consists of experts in cybersecurity, adversarial machine learning, and integrity in addition to multilingual content specialists with background in integrity issues in specific geographic markets.

Public data summary

Source: Meta The provider has not supplied this information.

Model Specifications

Context Length10000000

LicenseCustom

Training DataAugust 2024

Last UpdatedMarch 2026

Input TypeText,Image

Output TypeText

ProviderMeta

Languages12 Languages

Quick Start