God Mode Podcast

Quote · God Mode Podcast

Anthropic made Opus dumber (73% Drop), Claude Code Hits $2.5B ARR & Tesla AI5 | God Mode Pod EP09

Explore episode Apr 18, 2026

For now, it just feels like the providers have the upper hand because they are able to subsidize this, whereas Lovable, they can't subsidize. They're still paying for API credits straight to Anthropic.

Ben

Anthropic made Opus dumber (73% Drop), …

Where this was said

Claude Opus 73% Reasoning Drop: Nerfed for Compute

At 28:34 · chapter starts 25:40

Technology

Anthropic Quietly Nerfed Claude Opus by 73%

Anthropic made Opus dumber (73% Drop), Claude Code Hits $2.… · Apr 18, 2026 Technology

An AMD director analyzed 7,000 Claude Opus sessions and found a 73% drop in reasoning depth between January and March 2026. Anthropic didn't change the model — they throttled how much it thinks to save compute. Power users noticed.

Technology

73% drop in reasoning depth

Anthropic made Opus dumber (73% Drop), Claude Code Hits $2.… · Apr 18, 2026

An AMD director analysis of 7,000 Claude Opus sessions found a 73% reduction in reasoning depth between January and March — Anthropic quietly throttled thinking to conserve compute.

Similar quotes

I think young people will go on Discord more than they check their emails. So for step 5, another thing you want to do is turn all of your early users into potential advocates for your product.

Sam

I built a $14K/month SaaS using Discord…

I opened up a notepad and I blurted out on the ChatGPT voice mode my whole idea. So I was given a bunch of code and I was just copying it back and forth into notepad.

Sam

I built a $14K/month SaaS using Discord…

You need to initially bet on yourself, think that, okay, cool, I'm going to have 100,000 users building my tool. And you want to have the frameworks behind that and you might not know them, but when you tell Cursor to think that way, it will change its mindset in the way it prompts you to do things.

Sam

I built a $14K/month SaaS using Discord…

I thought, okay, why not spend a weekend to just try and figure out like what this AI thing is. The best way for me to learn is to build something with it.

Bhanu

The Marketing Playbook That Grew My App…

If I want to create a new tool, it takes less than 5 minutes for me to do it because I already have all these tools. I can just tell Cursor, okay, look at these free tools. Now in a similar way, create this new free tool.

Bhanu

The Marketing Playbook That Grew My App…

If you know your users are falling off during onboarding, then you know which screens aren't selling your app well enough. If you track all of the feature clicks on your app, you can identify your app's most compelling value proposition and then communicate that in your marketing.

Yali

Why Product Is Better Than Distribution…

A hard paywall is unskippable.

Guest

Two pricing strategies for mobile apps …

For these tool-focused apps like PuffCount, ads don't really work because you're not expecting the user to be on your app for a super long period of time.

Guest

Two pricing strategies for mobile apps …

In my first games, I monetized through ads because people were on the app all the time. They were playing the game a lot. So the ads worked well for Grid and Wordle.

Guest

Two pricing strategies for mobile apps …

Up next:

0:00