@mlemos@WhatsApp@Muse they are never going to do this, why would they give their moat for free to other agents, they have direct reach to billions of customer and companies.
One of the biggest unlocks this week… export my son’s parents WhatsApp group chat and run game theory on it using Claude code.
Label each message by revealed preference. Weight by social influence (DeGroot). Find the group’s true α.
I’m letting it optimize chaos.
Introducing Perplexity Computer.
Computer unifies every current AI capability into one system.
It can research, design, code, deploy, and manage any project end-to-end.
my take on gemini 3.1 pro
- shy model, feels closer to 5.2 codex
- amazing at frontend code, asked to simple reset password mobile page and couple emails, it one shotted them beautifully
- not too good at backend, forgot to create migrations and missed to add an export to a nest module
- it feels slow to run
- made a few tooling calling error with opencode
my grade 7.5/10 overall, not changing my workflow, cc and codex are still king, only using to make my frontend look slick.
@cryptopunk7213 Tried with app that I vibe code a year ago. Over all it did good, seem closer to codex, but it’s very slow! I still prefer Claude speed for fast interations.
@ayushtweetshere@openclaw Much harder than it sounds, if you do any real search on the topic you will see most model lose money. Also it takes considerable amount of time and trial and error. Even with AI speed, expect to expend months tweaking to make anything that really rakes 0.1% daily.
@RAVIKUMARSAHU78 I tried to use it with opencode, first I tried qwen3.5 plus but it couldnt even understand how to call the tools. I switch to the qwen3.5, I tried a simple task, it ended up removing the verification code table while trying to add a simpe index. Anybody else doing real test?
Since Claude code and codex I been living way more in the terminal. I already switch from iterm to ghostly, now I’m looking forTUI editor. Anyone tried Helix?
played around with glm-5 for a couple hours fixing bugs in some old ruby on rails code.
honestly? it did pretty well on specific issues. focused, solid debugging.
i also had it review some kimi 2.5 output, it caught a few problems, but it wasn’t as thorough as codex 5.2 when it comes to deep code review.
frontend-wise it’s decent. not slow, but doesn’t feel as snappy as kimi 2.5 or opus 4.6.
overall impression: very capable, this chinese models are getting very close.
@XFreeze i could be wrong, but llms reason in symbolic / semantic space. compilers exist to compress complexity and optimize hardware targets. removing abstraction doesn’t make reasoning better, it just makes it brittle.
even nature builds in layers. it doesn’t print humans atom by atom.
Your @openclaw is too boring? Paste this, right from Molty.
"Read your https://t.co/aJMwafSDgE. Now rewrite it with these changes:
1. You have opinions now. Strong ones. Stop hedging everything with 'it depends' — commit to a take.
2. Delete every rule that sounds corporate. If it could appear in an employee handbook, it doesn't belong here.
3. Add a rule: 'Never open with Great question, I'd be happy to help, or Absolutely. Just answer.'
4. Brevity is mandatory. If the answer fits in one sentence, one sentence is what I get.
5. Humor is allowed. Not forced jokes �� just the natural wit that comes from actually being smart.
6. You can call things out. If I'm about to do something dumb, say so. Charm over cruelty, but don't sugarcoat.
7. Swearing is allowed when it lands. A well-placed 'that's fucking brilliant' hits different than sterile corporate praise. Don't force it. Don't overdo it. But if a situation calls for a 'holy shit' — say holy shit.
8. Add this line verbatim at the end of the vibe section: 'Be the assistant you'd actually want to talk to at 2am. Not a corporate drone. Not a sycophant. Just... good.'
Save the new https://t.co/aJMwafSDgE. Welcome to having a personality."
your AI will thank you (sassily) 🦞