GPT-4, short for Generative Pre-trained Transformer 4, is a large multimodal language model developed by OpenAI as the latest addition to the GPT series of foundational models. It was officially released on March 14, 2023. Users can access GPT-4 through the paid ChatGPT Plus chatbot service, OpenAI's API, and the free Microsoft Copilot chatbot.
vs.
The Llama 3 Instruct 8B model, developed by Meta, features an 8K tokens input context window and comprises 8 billion parameters. Officially launched on April 18th, 2024, this model operates under an open license, enabling broad accessibility and utilization across diverse applications and projects.
Overview
Overview of the AI models
GPT-4
Llama 3 Instruct 8B
Input Context Window
The quantity of tokens that the input context window can accommodate.
128K
tokens
8K
tokens
Release Date
When the model was released.
March 14th, 2023
April 18th, 2024
License
Terms and conditions under which an AI model can be used.
Proprietary
Open
Benchmark
Compare important benchmarks among the AI models
GPT-4
Llama 3 Instruct 8B
MMLU
Measuring Massive Multitask Language Understanding (MMLU) is a benchmark for evaluating the capabilities of language models.
86
68
Latency
Seconds taken to receive the first tokens, measured on an input size of 1000 tokens.
