Building with Ai 👀 Learning Backend 💻
Earth
Joined September 2023
- Tweets434
- Following372
- Followers78
- Likes2.1K
a lot of people only pay for ChatGPT because of codex's separate limits. now they're gonna share the same limits?
that was literally the only thing keeping a lot of people from switching to Claude, and now you're giving them a reason to leave.
Amr Rady retweeted
OpenAI, I have only two words for you for merging ChatGPT and Codex usage limits:
SCREW YOU.
Amr Rady retweeted
🚨 KIMI K3.1 PRO IS LAZY WHEN COMPARED TO GEMINI 4 ARGON
I made Kimi K3.1 Pro (HIGH) and Gemini 4 Argon (HIGH) build a 3D game around the flying Tesla Roadster 🚗
Kimi took ~28 min and cost ~$0.92
Gemini took ~39 min and cost ~$1.65
Both ran through @MixRoute_ai using OpenCode on high effort.
Kimi's physics seem a bit more realistic, but Gemini built a MUCH bigger world with way more detail and better overall graphics.
Which one would you actually play?
And if you want to try building your own mini world, @MixRoute_ai is giving new users $5 in FREE API credits 👇
🤝 Paid partnership
Introducing ChatGPT's latest update: Failure-as-a-Service.
We've been working hard to bring you features nobody asked for.
Here's what's new:
Retry From Scratch
Spent 20 minutes researching something? Hit Retry and watch ChatGPT start the entire process again. Your progress was overrated anyway.
Ghost Messages
Send a message and watch it disappear like it never existed. Don't worry, it might randomly show up a few minutes later.
Connection Roulette
Your connection is fine. ChatGPT just decided it isn't.
Conversation Eject
Deep into an important task? Perfect time to tell you the conversation is too long and force you into a new chat.
Infinite Retry
Something went wrong? Retry. Failed again? Retry. Still broken? You know what to do.
And the best part? None of this existed 3–4 months ago.
Somehow we went from "using the product" to "being the QA team." Unpaid, of course.
@OpenAI @ChatGPT what happened to this product?
ChatGPT Work is back again!
they rolled it out without enough clarity, so they brought it back.
Amr Rady retweeted
I reverse engineered Resend, and it's 78% cheaper.
✨✨ Introducing reloop.sh ✨✨
Everything you need to send email from your app...
⚡ Email API
📮 SMTP relay
📨 Inboxes for your AI agents
🔁 Automations
🛡️ Deliverability guardrails
🔗 Bring your own SMTP (and keep your reputation)
And it's not a dinosaur tool:
+ It's open source; you can run it for free
+ It's agentic; everything in the UI is available to your AI
Move over from Resend in 2 minutes.
100k emails for $20/month, not $90. Free up to 3,000/month.
Oh, and it receives email too, not just sends it.
I vibe-coded beehiiv, and it's 83% cheaper.
✨✨ Introducing MailCheap.ai ✨✨
Everything you need to run a newsletter...
✍️ Editor
📅 Scheduling
📈 Stats
🌐 A website for every issue
🛡️ Deliverability guardrails
🔗 Send from your own domain (and keep your reputation)
And it's not a dinosaur tool:
+ It's open source; you can run it for free
+ It's agentic; everything in the UI is available to your AI
Import from beehiiv, Substack, or Mailchimp in 2 minutes.
You pay for the emails you send, not for dormant subscribers. From $19/month.
Oh, and no "Powered By" badge at the bottom of your emails.
When the task is finished, I change the model to Haiku 5.5 to compaction. It doesn't cost anything, or maybe only 1% of the 5h limits.
It works perfectly for me.
Claude Code just added idle compaction, and this could save a LOT of usage.
Imagine you have a session with 500K tokens of context. You come back 90 minutes later, your prompt cache has expired, and Claude has to process all 500K tokens again without the cache. That's insanely expensive.
Now, Claude automatically compacts idle sessions before the cache expires. Instead of reprocessing 500K tokens, you might only need to process, say, 100K.
And imagine the difference with an 800K-token session.
I always recommend manually compacting after finishing parts of a task, but most people never do it.
This feature could save the average Claude Code user a ton of usage.
Disk space improvements are great, but what about RAM?
I have ZERO active sessions, yet the ChatGPT macOS app is using 8.35 GB of RAM across 80 helper processes.
That's more than half the memory on my Mac, while doing literally nothing.
Can we get some memory optimizations too?
I was recording a voice message in Grok Bot when the update screen suddenly popped up and interrupted me.
Everything I'd recorded was lost. There was no way to go back or recover it. The only option was to restart.
@bot please fix this. Updates shouldn't interrupt an active recording and wipe it out.
