Model reference · open weights

Llama-2

LLMs meta-llama Text gen 3 builds Open, with conditions 804k dl/mo

Llama-2 is an open-weight language model from Meta. Llama-2-70b-chat-hf (FP16) weighs 138 GB; the smallest configuration that runs it is B200 180 GB.

Llama-2 is a 6.7B parameter generative text model developed by Meta for English text generation tasks.
It utilizes an optimized transformer architecture and is distributed under the llama2 license.
The model supports a context length of 4k tokens and is part of a family of models ranging from 7B to 70B parameters.

Summary of the meta-llama/Llama-2-7b-hf model card, 2026-10-01. The estimate below is for another build of the family.

What it is

What it runs on

Memory and cards for Llama-2-70b-chat-hf (FP16)

138 GBweights, file size

762 MBruntime overhead, at least

How much memory each request adds isn't estimated yet for this architecture. The weights need at least the cards below, plus room for the context.

Builds

Build	Params	Precision	Weights	Smallest setup
Llama-2-7b-hf ↗	6.7B	FP16	13.5 GB	RTX 4060 Ti 16 GB
Llama-2-13b-chat-hf ↗	13.0B	FP16	26.0 GB	RTX 5090 32 GB
Llama-2-70b-chat-hf (above) ↗	69.0B	FP16	138 GB	B200 180 GB

How it works

Details

Languages