Llama 3.3 70B Instruct

Conditional

Meta · Llama · Released 2024-12

70BText

A late-2024 refresh that gets close to Llama 3.1 405B quality on many benchmarks at a fraction of the compute cost.

Strengths

  • +Near-405B quality on many text benchmarks
  • +Improved instruction following and multilingual support
  • +128K context window

Limitations

  • -Text-only, no native vision support
  • -Still needs a multi-GPU or heavily quantized setup to self-host

License

Llama 3.3 Community License

free with acceptable-use or scale conditions - read it

Hardware

wants 24-48GB of VRAM - a 3090/4090-class card or better

Links

WeightsGitHub

Stats

meta-llama/Llama-3.3-70B-Instruct— downloads·— likes

via Hugging Face

Related models

Llama 3.1 8B Instruct

Meta · Llama

Llama 3.1 70B Instruct

Meta · Llama

Llama 3.1 405B Instruct

Meta · Llama

Llama 3.2 3B Instruct

Meta · Llama