Sean Murphy

Turning bold ideas into things that actually work. AI strategy, systems thinking, and execution-first consulting for founders and builders.

[Sean Murphy @AmbRealismX]

It's a pretty cool feeling to release my first OSS project forked from one person that influenced me: @theo to include another incredable project from another who influenced me @badlogicgames and then use my "quick" reflexes to capture a repost from the creator of Pi!

[Sean Murphy @AmbRealismX]

So good…

[Love Music @khnh80044]

A stunning rendition of "Dreams" by Fleetwood Mac, beautifully performed by Lanie Gardner ❤️🎶✨

[Sean Murphy @AmbRealismX]

Yesterday I did three long tail tests pitting GLM 5.2 against Opus 4.8 and GPT 5.5. Two of the tests used Dynamic Workflows (Claude Code for Opus, Pi with the dynamic workflows extension for GLM and GPT) and the third was a “/goal” task. Let’s just say that GLM-5.2 took 2 out of three and Codex took the other. Opus came in last in all three.

[Kevin Bass @kevinnbass]

. @AnthropicAI released Fable 5 and told us it would switch to API-only for Fable 5 on June 22.

A lot of us purchased additional Max subscriptions to capitalize before the 22nd.

Anthropic knew we would do this.

Then it rug-pulled and took away Fable access almost as soon was released.

Anthropic reset our weekly limits after the rug pull, but that's not enough.

We cannot even use the API spend on these accounts because of how much Anthropic throttles the API on Max accounts that use Opus 4.8.

So now these accounts are just sitting around. Paid for with money countless customers burned because they were misled.

There is a simple solution to this. It is not to reset the weekly limits.

We paid money to get something that we did not receive.

What we did receive is worthless.

We paid Anthropic money, only to receive nothing.

Anthropic owes its customers refunds.

[Sean Murphy @AmbRealismX]

@djtigon @aditiitwt M3 is good but GLM-5.2 for code review and planning when on Max is on par and sometimes beats GPT-5.5 and Opus 4.8. I spent most of the day testing it against 5.8 and Opus and it really performed well.

[Sean Murphy @AmbRealismX]

@Moustache_bq @KillaKreww You’re just arriving at that now in 2026? Welcome to earth.

[Sean Murphy @AmbRealismX]

@DavidOndrej1 @DissentFu Don’t be an a$$hole

[Sean Murphy @AmbRealismX]

@Gc_qube

[Sean Murphy @AmbRealismX]

@seanarnould1 @AlexFinn Oh man. You haven’t figured it out yet have you?

[Sean Murphy @AmbRealismX]

@yoemsri @AlexFinn This ^

[Sean Murphy @AmbRealismX]

Anyone else tired of The Formula?

“OMG, latest frontier model one-shotted a clone of someone else’s hard work!”

Then: “I BUILT IT in X prompts.”

Cool. Another Rubik’s Cube sim. Another Lovable clone. Another Minecraft clone.

What if we built on OSS, or made something new?

[Sean Murphy @AmbRealismX]

@rileybrown @daytonaio So many people VibeXeroxing. 🙄

[Sean Murphy @AmbRealismX]

@CtrlAltDwayne I don’t understand why with such a capable model everyone is using it like a Xerox.

[Sean Murphy @AmbRealismX]

@Dussyme 4

[Sean Murphy @AmbRealismX]

@medoliyy 40

[Sean Murphy @AmbRealismX]

@petergyang @agents Have you been in a coma for a year?

[Sean Murphy @AmbRealismX]

@OpenAI @grok explain the video

[Sean Murphy @AmbRealismX]

@KanikaBK By new do you mean 16 months ago or did you go back in time when you posted this?

[Sean Murphy @AmbRealismX]

Question for @OpenAI: why keep users blind to the context window? Give us a real meter: tokens in, tokens left, and what will be dropped next. Serious users manage budgets, files, and risk. Hiding the window makes the tool feel less trustworthy. @sama