lactroiii/Llama-3.3-70B-Instruct

TEXT GENERATIONConcurrent Unit Cost:4Model Size:70BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 8, 2026License:llama3.3Architecture:Transformer Featherless Exclusive Cold

lactroiii/Llama-3.3-70B-Instruct is a 70 billion parameter instruction-tuned generative language model developed by Meta, based on the Llama 3.3 architecture. Optimized for multilingual dialogue use cases, it supports 8 languages and utilizes Grouped-Query Attention (GQA) for improved inference scalability. This model excels in assistant-like chat applications and outperforms many open-source and closed chat models on common industry benchmarks.

Loading preview...

Model Overview

lactroiii/Llama-3.3-70B-Instruct is a 70 billion parameter instruction-tuned generative language model from Meta, built on the Llama 3.3 architecture. It is optimized for multilingual dialogue and designed for commercial and research use. The model was pretrained on over 15 trillion tokens of publicly available data with a knowledge cutoff of December 2023, and fine-tuned using supervised fine-tuning (SFT) and reinforcement learning with human feedback (RLHF).

Key Capabilities

  • Multilingual Support: Optimized for 8 languages including English, German, French, Italian, Portuguese, Hindi, Spanish, and Thai, with potential for fine-tuning in others.
  • Optimized Architecture: Uses an optimized transformer architecture with Grouped-Query Attention (GQA) for enhanced inference scalability.
  • Instruction-Tuned: Specifically designed for assistant-like chat and dialogue use cases.
  • Tool Use: Supports multiple tool use formats, enabling integration with external functions and services.
  • Performance: Achieves strong results on benchmarks such as MMLU (86.0), HumanEval (88.4), and MATH (77.0), outperforming previous Llama 3.1 models in its class.

Intended Use Cases

  • Assistant-like Chat: Ideal for building conversational AI agents and chatbots.
  • Multilingual Applications: Suitable for applications requiring understanding and generation in supported languages.
  • Research and Commercial Use: Intended for a wide range of commercial and research applications, including synthetic data generation and model distillation.
  • Tool Integration: Can be used in systems requiring function calling and interaction with external tools.

Finetunes
2 models