ChatGPT Routing Broken? Why I Switched Models

Next Tech and AI · Intermediate ·🧠 Large Language Models ·3mo ago
I cancelled my ChatGPT (GPT-5) subscription after 10 months — here's why and which AI I'm switching to. After months of solid performance, GPT-5 became inconsistent for real work: routing problems, declining output quality even in "thinking" mode, and regressions in things that used to work reliably. In this video I show concrete examples, head-to-head tests with Claude and Google Gemini, and the practical alternatives I'm evaluating for day-to-day work. This is my unsponsored, hands-on take (tested on real projects, not benchmarks). — What I cover — • Why GPT-5's routing is a “black box” pro…
Watch on YouTube ↗ (saves to browser)

Chapters (7)

ChatGPT Subscription cancelled
0:24 What worked well with ChatGPT
2:12 Inconsistent Quality
4:35 The Black Box — Routing and hidden settings
7:01 Features no longer exclusive
8:07 GPT-5.1 — the answer?
10:20 The new 'Main' LLM
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)