---
title: "GPT-4 — LLM profile"
provider: "OpenAI"
version: "gpt-4-0613"
model_type: "llm"
context_window: "8192"
parameters: "1760B"
url: "https://lforla.org/models/gpt-4"
markdown_twin: "https://lforla.org/markdown/models/gpt-4.md"
---
# GPT-4

## Overview

GPT-4 is OpenAI's flagship large-scale multimodal language model, released in March 2023. It accepts both image and text inputs and produces text outputs, achieving human-level performance on a wide range of professional and academic benchmarks, including passing the simulated bar exam in the top 10% of test takers.

## Capabilities

- Multimodal (image + text inputs, text outputs)
- Strong performance in reasoning, coding, and writing
- 32,768-token context window (base model)
- Robust to adversarial prompting and safety evaluations

## Technical details

- Developed with a transformer-based architecture and large-scale alignment
- Uses RLHF (Reinforcement Learning from Human Feedback)
- Trained to predict the next token and optimized for helpfulness and safety

## References

- GPT-4 Technical Report — [arXiv:2303.08774](https://arxiv.org/abs/2303.08774)