Build on your own terms

Optimized models for easy deployment, cost efficiency, and performance that scale to billions of users.

Latest Llama models

The latest models feature native multimodality, advanced reasoning, and industry-leading context windows.

Llama 4

Native multimodality leveraging early fusion to pre-train unlabeled text and vision data enabling a change in intelligence from separate, frozen multimodal weights.

  • 10M-token context for long-form work
  • Multimodal text + image understanding
  • For use cases around memory, personalization, and multi-modal applications

Llama 4 Maverick

  • Natively multimodal for image and text understanding.

More details
Download models

Llama 4 Scout

  • Offers single H100 GPU efficiency
  • 10M context window
  • For use cases around long document analysis
    Download models

Llama 3

The open-source AI models you can fine-tune, distill and deploy anywhere. Choose from our collection of models: Llama 3.1, Llama 3.2, Llama 3.3.

  • Llama 3.3: Multilingual open source large language model.

    • Available in 70B
    • Experience 405B performance and quality at a fraction of the cost
    • Built for text-based use cases such as synthetic data generation
      More details
      Download models

  • Llama 3.2: Flexible and cost-effective, built for edge use cases.

    • 1B & 3B lightweight and cost-efficient
    • 11B & 90B multimodal models for reasoning on high resolution images

More details
Download models

  • Llama 3.1: Open-foundation model built for flexibility and control.
    • Available in 8B, 70B, and 405B sizes
    • Capabilities in general knowledge, steerability, math, tool use, and multilingual translation

More details
Download models

Model optimization

  • Prompt Engineering: Used to improve performance of LLMs.
    Learn more
  • Fine-tuning: Adapting pre-trained models for specific use cases.
    Learn more
  • Vision capabilities: Understanding and reasoning over images and text.

Learn more

  • Quantization: Reducing computational and memory requirements.

Learn more

  • Distillation: Teaching smaller models to match a larger model's performance.
    Learn more
  • Evaluations: Tests to systematically measure performance.
    Learn more

Llama 4 capabilities

  • Natively multimodal
  • Unparalleled long context
  • Expert image grounding

Llama 4 benchmark

Task Metric Llama 4 Maverick Llama 4 Scout
Reasoning MMLU Pro 80.5 74.3
Coding LiveCodeBench 43.4 32.8
Multimodal (Image) MMMU 73.4 69.4
Multilingual MMLU Multi 84.6 74.3
Efficiency Cost per 1M tokens $0.19–$0.49 $0.19–$0.49

Featured case studies

How Stoque is using Llama

Transforming internal intelligence with Llama. Stoque enabled teams to find insights faster, reduce friction, and work more efficiently at scale.

  • 50% reduction in repetitive queries for technical support
  • 30% more administrative and support tasks completed
  • 11% increase in internal user satisfaction
    Read more

How Shopify is using Llama

Shopify uses Llama to generate product pages, localize content, and automate support.

  • +76% higher token throughput than the previous model
  • 97.7% accurate Macro-F1 score on intent detection
  • 33% compute cost savings with JSON output
    Read more

Protections in the era of generative AI

Comprehensive system-level protections proactively identify and mitigate potential risks, empowering developers to more easily deploy generative AI responsibly.