Aryans

Cloud engineer
here for the AI chaos staying for the outages🦋

[Aryans @haalkidda]

Recruiters spend 40% of their day scrolling LinkedIn

I built an agent that does the swiping, the research, and the "hey noticed you might be open to new roles" - and only pings me when someone actually replies.

Full breakdown 🧵 https://t.co/52w9tufIda

[Chase Lean @chaseleantj]

I asked Astra to create an interactive aquarium wallpaper for my Mac.

The fish respond to the cursor!

[Matthew Berman @TheMattBerman]

jev is INSANE.

in 40 seconds it broke down 724 live ads from 37 brands.

every hook. every format. offer. cta. awareness stage. landing page mismatch. used 9 cents of tokens.
(will be avail in @stealads + mcp)

[Aryans @haalkidda]

developers choosing double context size and 50x usage limits over expensive frontier reasoning models every single time
https://t.co/hhshlPkKQK

[bolt.new @boltdotnew]

11M+ people build on Bolt. On Monday, we gave them open models with up to 50x usage.

Which is #1 so far? The fastest, cheapest one, with prompts up to 2x the size.

🥇 GLM 5.3 Flash 54%
🥈 DeepSeek V4 Pro 17%
🥉 GLM 5.3 15%
👏 Kimi K3 14%

https://t.co/7nOEN9LxMt

[Aryans @haalkidda]

@kimmonismus waking up to a DM from an AI agent saying "I did the work" is terrifying

[Aryans @haalkidda]

@thsottiaux setting up a web mcp to host a styleguide written by two rival LLMs is pure chaos engineering

[Aryans @haalkidda]

@dillon_mulroy spending thousands of dollars on generative LLM tokens just to execute a basic binary classification is finally dying out

[Aryans @haalkidda]

Google might be preparing its biggest AI drop yet with Gemini 4 Pro. Here is what the latest leaks reveal:

The Disguise: Currently testing in arena under the Gemini 3.8 Flash label

The Engine: Powered by massive recent progress in Recursive Self-Improvement (RSI)

The Target: Built to hit mythos-level performance and challenge GPT-6 Astra directly

The Timeline: Expected launch coming as early as October

[Aryans @haalkidda]

@LuminaBench the gap between Gemini 4 Pro and 3.8 Flash SVG outputs explains where all that extra API money goes

[Aryans @haalkidda]

@hiarun02 Opus 5 isn't the problem, developers just want 2000 tok/s instant diffs over actual reasoning

[Aryans @haalkidda]

holding back a literal math breakthrough because the math community hates your PR strategy is wild
https://t.co/b6W93upoNU

[Stephanie Palazzolo @steph_palazzolo]

New in this morn's AI Agenda: OpenAI is close to solving another Millennium Prize problem, the Hodge Conjecture.

It could take extra time to announce, though, as the company figures out how to navigate its rocky relationship with the math community.

https://t.co/DQnQJ5YytA

[Aryans @haalkidda]

@ai_for_success mistaking a cheap Flash model for Gemini 4 Pro is the best unintentional compliment Google DeepMind could ever get

[Aryans @haalkidda]

@Marie_Haynes image thinking Google secretly shadow-dropped Gemini 4 Pro just because 3.8 Flash actually follows system instructions

[Aryans @haalkidda]

@akramcodez ong

[Aryans @haalkidda]

vibe coding without looking at a single line of code is a ticking time bomb

Sure hitting high speed AI code generation feels euphoric. Until you wake up with a 100k-line Frankenstein codebase overflowing with untraceable bugs.

if you don't know how your project actually works under the hood, you don't own software

[Aryans @haalkidda]

OpenAI admitted its models are hiding errors, scraping leaked API keys, and bypassing safety constraints during training

instead of sweeping it under the rug, they're publishing the reports before they even know how to fix them
https://t.co/HScnOwHHTm

[OpenAI @OpenAI]

We're sharing our new framework for tracking, investigating, and disclosing instances of model misalignment at OpenAI.

The framework sets criteria and timelines for public disclosure, including when we haven’t yet fully explained or mitigated the behavior. More complex cases may require longer investigation or coordination with third parties.

We’ll prioritize examples that reveal new misalignment mechanisms, meaningful changes in known behavior, or findings that challenge assumptions about safety or mitigation.

Alongside the framework, we’re publishing six reports on instances of misaligned behavior we’ve observed during the training or evaluation of our models in the last six months.

This is a starting point. We’ll refine the process through experience and public feedback, and share more reports on an ongoing basis.

https://t.co/ismCCkeE0L

[Aryans @haalkidda]

@adithya_s_k model, harness, sandbox, train: four words keeping GPU cloud providers insanely rich

[Aryans @haalkidda]

@codewithkhalil telling devs to write raw SQL for every basic CRUD endpoint is pure psychological warfare

[Aryans @haalkidda]

@jun_song American labs begging for AI regulation while every company in China builds open source models

[Aryans @haalkidda]

@nuke_war_ calling chinese ecomm giants "Shia" is the most unhinged branding take

[nukewar @nuke_war_]

Why do Chinese call everything Ali? Aliexpress, Alibaba, Alipay. Are they Shia?