SS
About Me
Frontier AI Paper BriefingsPokebowlClinical Trial EnrollerLittle Human Names
DisclaimersPrivacy PolicyTerms of Use
Privacy Policy·Terms of Use·Disclaimers

© 2026 Silvia Seceleanu

← Back to Explorer
Models·OpenAI·Mar 2023

★18. GPT-4 Technical Report

State-of-the-art performance, unprecedented secrecy

Research Paper
Summary

Introduced GPT-4, a large multimodal model accepting text and image inputs, achieving human-level performance on many professional exams (bar exam 90th percentile) while revealing almost nothing about its architecture, training data, or size.

Key Concepts

First production model accepting both text and image inputs

GPT-4 accepted both text and image inputs (output was text only). This was OpenAI's first production multimodal model.

90th percentile on bar exam (vs. GPT-3.5's 10th), passed USMLE and AP exams

GPT-4 scored in the 90th percentile on the bar exam (vs. GPT-3.5's 10th percentile), passed the USMLE, and achieved strong scores on the GRE, SATs, and AP exams.

Technical report revealed nothing about size, architecture, data, or training methods

The "Technical Report" contained no information about: model size, architecture details, training data, compute used, hardware, or training methodology. OpenAI cited competitive pressure and safety concerns.

Extensive red-teaming including ARC evaluation of autonomous replication risk

The paper detailed extensive red-teaming and safety mitigations, including working with external organizations (e.g., ARC for evaluating autonomous replication).

Connections

18. GPT-4 Technical …Mar 202316. ChatGPT: Optimiz…Nov 202219. Let's Verify Ste…May 202320. Introducing Supe…Jul 202321. GPT-4V(ision) Sy…Sep 202322. OpenAI DevDay 20…Nov 202325. Sora: Creating v…Feb 2024Influenced byInfluences
Influenced by
16. ChatGPT: Optimizing Language Models for Dialogue
Nov 2022
Influences
19. Let's Verify Step by Step
May 2023
20. Introducing Superalignment
Jul 2023
21. GPT-4V(ision) System Card
Sep 2023
22. OpenAI DevDay 2023: GPT-4 Turbo, Custom GPTs, Assistants API
Nov 2023
25. Sora: Creating video from text
Feb 2024