Llama 3.3 70B

The Meta Llama 3.3 70B Instruct multilingual model is an instruction tuned generative model in 70B (text in/text out). Privatemode provides a variant of this model that was quantized using AutoAWQ from FP16 down to INT4 using GEMM kernels, with zero-point quantization and a group size of 128.

Model ID

llama-3.3-70b

Source

Hugging face

Modality

• Input: text
• Output: text

Features

• Streaming
• Tool calling
• Structured outputs

Context limit

• Context window: 70k tokens
• Max output length: 4028

Endpoints

• /v1/chat/completions
• /v1/completions