Is GPT-5.5 Better Than Opus Now? (ft. Our New AI Co-Host) - EP99.38

Join Simtheory: https://simtheory.ai

So Chris, this week we finally give our GPT-5.5 impressions (it's actually great), introduce our new AI co-host Moshi (who immediately embarrasses himself), argue about whether the OpenAI/Jony Ive phone is genius or doomed, witness Grok 4.3's unhinged infinite emoji meltdown, declare Opus 4.7 the first-ever Anthropic regression, get excited about GPT Real-Time Voice 2.0 as the future of agentic workflows, debate whether token prices will ever come down, and play the worst diss track in show history. Watch my spud.

CHAPTERS:

0:00 - Intro & Introducing Our New AI Co-Host Moshi

1:39 - Trying to Break Moshi: The Illegal Cigarette Trade Test

2:30 - OpenAI's Jony Ive Phone: Do We Need a Device?

5:07 - Telegram Agents & GPT Real-Time Voice 2.0 Dream

7:38 - The Supervisory Agent: Managing Your Agentic Workflow

9:05 - Wait... Are We Accidentally Validating the OpenAI Phone?

11:37 - GPT-5.5 First Impressions: Actually Really Good

14:36 - 5.5 vs Opus 4.6: Different Strengths

17:00 - Opus 4.7: The First-Ever Anthropic Regression