GPT-5.4 Is Here (Worse Than Sonnet 4.6?)

Ryan & Matt Data Science · Beginner ·🧠 Large Language Models ·3w ago
Want to learn how to use AI? Join our skool group, we also have a free weekly call: https://www.skool.com/data-and-ai GPT-5.4 just dropped and OpenAI is calling it their most capable model yet — built for "professional work." But does it actually deliver? In this video, I break down the GPT-5.4 press release, benchmarks, pricing, and then put it head-to-head against Claude Sonnet 4.6 with real-world tests: niche knowledge questions, spreadsheet data cleanup, and more. The results might surprise you. We compare API pricing across GPT-5.4, GPT-5.4 Pro, Claude Sonnet 4.6, Claude Opus, and Gemi…
Watch on YouTube ↗ (saves to browser)

Chapters (15)

GPT-5.4 Is Here
0:23 Press Release Breakdown
1:00 Benchmarks Overview
1:21 Knowledge Work & Spreadsheet Claims
1:55 Computer Use & Vision
2:34 Availability & Pricing Breakdown
3:33 GPT-5.4 vs Claude vs Gemini Pricing Comparison
4:32 Test 1: Niche Knowledge — Most Valuable Baseball Cards
5:45 Test 1 Results: Sonnet vs GPT-5.4
6:33 Test 2: Niche Knowledge — Oldest Presidential Card Sets
7:40 GPT-5.4 Hallucinations vs Sonnet Accuracy
9:00 Test 3: Spreadsheet Data Cleanup
10:10 Sonnet Spreadsheet Results
10:50 GPT-5.4 Spreadsheet Results
11:20 Final Verdict & Takeaways
5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
Next Up
5 Levels of AI Agents - From Simple LLM Calls to Multi-Agent Systems
Dave Ebbelaar (LLM Eng)