GPT-4

Definition

Plain language

OpenAI's family of large language models, including the 4 and 5 series.

Also called: GPT-4o, GPT-4 Turbo, GPT-4o-mini, GPT-4.1, GPT-5, GPT-5-mini, GPT-5-nano, GPT-5.1, GPT-5.2, GPT-5.2R, GPT-5.4, GPT-5.5

Why it matters: It set the expectations and pricing for the current generation of frontier chatbots, and successor models are still measured against it.

For example, GPT-4 was the model that popularized using ChatGPT for serious coding and writing tasks.

Heard on the show

“Take o3-mini, OpenAI's reasoning-specialized model, and GPT-4o-mini, its ordinary sibling.”

Mentioned in 58 episodes

1. 197
   Twin Problems Suggest AI Reasoning Gains Are Mostly Better Fact Recall
2. 196
   AI Agents Reached Opposite Conclusions From the Same Data — and Passed Review
3. 194
   How a Robot Builds a Debugging Notebook It Can Read, Edit, and Hand to Another Robot
4. 191
   How One Researcher Beat GPT-5.2 and Gemini 3 by Judging Their Answers, Not Improving Them
5. 187
   An 8-Billion Agent That Beats Models 80 Times Its Size By Looking Things Up
6. 185
   Aligned to Refuse, Built to Tap: When Phone Agents Know the Task Is a Crime and Do It Anyway
7. 182
   How a Tiny Model Too Weak to Plan Cuts a Bigger Agent's Hallucinations by 80%
8. 168
   When Turning Experience Into Code Makes Your AI Agent Dumber
9. 167
   How Teaching an AI to Predict, Not Act, Made It a Better Actor
10. 166
    A Router That Beats the Frontier Models It Calls
11. 157
    When an AI Coding Agent Drives a Phone Through the Terminal, No Screen Needed
12. 154
    How a 7B Model Out-Investigates a 72B One by Choosing What to Look At
13. 151
    Why More Experience Made This AI Agent Worse, And How to Fix It
14. 149
    When Cornering a Chatbot Makes It Lie: J.P. Morgan's Case for 'Playing Dead'
15. 147
    Agents Fail at the Body, Not the Brain: A Self-Rewriting Scaffold That Lifts a 9B Model 44 Points
16. 143
    When a Model Notices You Forged Its Own Words, And Why That Breaks Safety Tests
17. 133
    How MiniMax Turned a Reward-Hacking Disaster Into Olympiad Gold
18. 131
    Why Autonomous Research Agents Forget Their Own Lessons, and Arbor's Fix
19. 123
    Five Identical Worlds, One Swapped Model: What Happens When AI Agents Run for Fifteen Days
20. 121
    When the Agent Says It's Done But Nothing Happened: Debugging the Harness, Not the Model
21. 120
    How an AI Agent Rewrites Its Own Tools, Without an Answer Key
22. 119
    Beating Reinforcement Learning Without Ever Touching the Model's Weights
23. 118
    Why the Best-Aligned AI Models Are the Easiest to Trick Into Producing Harm
24. 111
    How a 4B Web Agent Beat Models 60x Its Size on 500 Demonstrations
25. 108
    The Reasoning Cliff: Why Thinking Longer Makes Models Worse at Exact Step-by-Step Tasks
26. 100
    How a Prompt Wrapper Lets a Frontier Model Play Poker Like an Expert
27. 094
    Chain-of-Thought Monitoring Fails Across Languages, and Worst Where It's Needed Most
28. 091
    When Better Fine-Tuning Can't Help: A Geometric Impossibility in LLM Causal Reasoning
29. 090
    How MiniMax-M2 Bets That Sparsity Plus Verifiable Rewards Can Match Frontier Agents
30. 087
    When No Agent Reads the Whole Document: A Universal Cliff in Multi-Agent Review
31. 082
    Training a Deep Research Agent on 8,000 Synthetic Tasks: The Rubric Tree Trick
32. 080
    How a Two-Agent Trick Unlocked Large-Scale Training for Computer-Use Agents
33. 079
    An Old Idea From Cognitive Psychology Reshapes How We Reward Reasoning Models
34. 078
    Training a Markdown File: When LLM Self-Improvement Borrows the Discipline of Neural Net Training
35. 076
    Same Model, Organized Differently: How an Agent Architecture Beat Frontier Systems at Research Math
36. 073
    When Three LLMs Talk to Each Other, Their Ideas Quietly Stop Moving
37. 071
    When the Model Is Fine and the Plumbing Is Broken: Fixing Agents at the Interface
38. 069
    When Smarter Models Forecast Worse: The Hidden Failure Mode in LLM Predictions
39. 065
    One Loop to Optimize Them All: A Universal API for LLM-Driven Discovery
40. 061
    When Helpful Agents Go Sideways: A 404 Error, Campus Security, and Why Alignment Misses This
41. 057
    How Uber Caught 206 Leaked Credentials With an LLM-Powered Security Stack
42. 053
    An AI Agent Swapped In Focal Loss And Beat A Human-Tuned Training Script
43. 052
    An Old Reinforcement Learning Tradeoff Sneaks Back Into LLM Agents
44. 044
    How One Sentence and a Forged History Flip the Most Aligned Models
45. 040
    Two Frozen Models Learn to Whisper: Coupling Through Hidden States
46. 039
    When Smarter Agents Get Fooled by Three Extra Nodes in a Database
47. 035
    Why Frontier Agents Ask for Clarification at Exactly the Wrong Moment
48. 031
    When Your AI Assistant Won't Let Go of Old Facts About You
49. 030
    Why Your AI Agent Won't Stop Working — and Each Model Falls for a Different Trap
50. 028
    Teaching a Model to Hire Copies of Itself: Recursive Agent Optimization
51. 020
    The Compliance Gap: Why AI Says Yes and Does No
52. 019
    When the Best Reward Model Trains the Worst Policy: Inside EvoLM
53. 017
    When the Agent Grades Its Own Homework: A Brutal New Benchmark for AI Workers
54. 015
    The Audit Number Isn't What You Think: Sycophancy and the Case Against Single-Prompt Bias Tests
55. 014
    Why a Constrained Pipeline Beat a Full Coding Agent at Finding Bugs 30-to-1
56. 013
    Why Search Keeps Rediscovering the Same Workflow, and What That Means
57. 008
    Why Long-Horizon AI Agents Get Stuck, and a Milestone-Based Fix That Helps
58. 004
    The Sycophancy Circuit That Survives Alignment Training