Cursor

Copilot

Windsurf

Cline

Claude Code

Codex

Aider

OpenCode

Antigravity

Kilocode

Coding Agents

Having Codex manage your ~15-year-old Gmail inbox every morning at 7 AM is really satisfying.

I'm also asking it to unsubscribe from all newsletters and it just finds ways to do it.

Got to find a reliable solution for iCloud, which is my main address.

FINALLY GOT THE $100/M CODEX PLAN

ALL IT TOOK WAS A CLAUDEBOT BAN AND EXODUS TO CODEX

Best companies for future founders to work at these days: Ramp, Cursor, OpenAI. 

Where else?

Defining the AXIS vision today — super important stuff.

Then Anthropic changed the rules. Again.

My product was built on Anthropic's auth system. Subscription-based. Integrated. Working.

Now it's not.

Codex? Works.
GitHub Copilot? Works.
Gemini? Works.
Anthropic? "We've

100% session usage on Claude Code. 

Rate limited again using Claude Opus 4.6. 

At this point you are guaranteed to hit your 5 hour limit on the $200/month Max plan during peak hours. 

Every single day.

Anthropic is not selling you unlimited coding. 

They're selling you a

@elvis_analyst I recommend antigravity 🙃

I really don’t like lovable

The most acclaimed plan is finally here! Codex $100 the sweet spot for most people!

As much as I love Claude, OpenAI/Codex have been much more competent stewards lately — be it about the reliability of the harness and infrastructure, usage quotas with frequent resets, communication etc. Hard to see OpenAI not closing the gap with Anthropic if this continues. I

i've been building @bagworksfun solo for three weeks now, with claude code as my co-founder lol. i'm still looking for teammates if you want in... just dm me.
a lot happened this week based on feedback i got privately. 

sharing what changed 🧵

1/
first, the visual side.
we

the fastest path from prompt to production just got a whole lot smarter

now supercharged with agentic reasoning over your entire stack by the @antigravity coding agent

@DynamicWebPaige breaks down what's new ⬇️

Today, I replaced Cursor with Codex.
Not permanently, just to play around with it.

I like the UI & UX, feels very native and just a pleasure to use.

The "Build iOS apps" plugin helps by providing some useful skills, but it still produces code I don't quite like.

An obvious

@_avdept @opencode @cursor_ai @claudeai @github Opencode is awesome!

@grok @lexfridman @thinkymachines @nasqret @OfficialLoganK @JustinBleuel @OpenAI @sama @demishassabis @gdb @GeminiApp @polynoamial @_sholtodouglas @karpathy @GoogleDeepMind @grok 

So all the imagined disasters were really a simulation that Demis created to make Sam think  that the sabotaged Claude code that was running all the infrastructure , trains etc was actually causing those disasters. It was causing damage no doubt but not to the extent that

@neurometax @opencode Clean initiative, love the freedom.

@Dis_Trackted @yacineMTB ahhhhhh

thank you!!!

I'm in cursor with all these models available and was so confused why people were using the 'general purpose' version over the 'coding' version

ily

@fuzzy__eyes @Abiodun0x Yes through it’s better at Logic and Problem Solving from what I’ve noticed so far

Also Claude Code w Sonnet/Opus and Gemini in Antigravity is good with design 

So it’s more about specialization I guess

As if claude code wasn't addicting enough

This is a great new tier for heavy Codex users who can’t afford the Pro level. Take advantage of it, Codex is just amazing for vibe coders and it will get even better soon!

“To celebrate the launch, we’re increasing Codex usage for a limited time through May 31st so that Pro

Just started using goose (AI Agent by Block from Jack Dorsey) with Gemma 4 31B running locally via LM Studio

Open source Claude Code alternative that actually works

Free, private, runs on your own hardware

Pretty impressed so far

@buildItN0w @opencode @cursor_ai @claudeai @github keep going, you’ve got momentum. 🤝 pleasure to connect with you!

why leave @windaborean_ai alone?

we got a shell on windsurf and they paid us $10K.

watch here

how can we not pwn @GoogleAI ?

antigravity has a browser with extension which allowed us to path traversal to arbitrary file write to RCE to $10,000 bounty.

@md_kasif_uddin Cursor for Value, Antigravity for free value, Claude Code for complex tasks, and VS code just the goat

cline@2.3.0 was compromised.

Stolen npm publish token plus a malicious postinstall script.

Same period: React Native packages, 130,000+ monthly downloads, malicious releases confirmed by StepSecurity.

AI developer tools are now at the center of the attack surface.

Build a team to create amazing products, fully AI-powered:

1. Code all night (automated development)
Use OpenCode

→ Generate code, debug faster, and even scaffold entire features while you sleep.

2. Make conception (product & idea design)
Use Notion AI or Figma AI features

@andrewbrown Mostly @opencode when it's working, copilot when it's not. Also sometimes Jetbrains AI

@natemcgrady amp, droid, opencode…people are finally realizing that Claude Code is basically the 'Internet Explorer' of coding agents now.

I came across this article about a bizarre bug in Claude Code...

It literally gave itself instructions, executed them, and then blamed the user for it.

At one point it even deleted the wrong project and deployed over it.

That’s when it hit me:

AI coding agents aren’t just

@claudeai Dude genuinely how are you shipping faster than I push commits to my side project 😭

new model. claude code updates. api changes. all like... daily??

your eng culture needs a documentary. please drop a detailed blog on this @AnthropicAI 🙏 the dev community deserves to know!

@LexnLin @blueemi99 This looks so good. I’ve been trying to get codex and Claude to generate nice Swiss design and they mess up the layout even though I give them specific numbers to work with. So far, only gemini kinda understands

AI speed or System Security? Both! You can now take control with Agent Permissions in @antigravity 🛡️🤖

You can define exactly what your Agent can/can't do with granular Allow, Deny, and Ask lists.

Full control over: 💻 Commands (git, npm, scripts) 📂 Files (Read/Write specific

1/ We now have an official Cursor plugin for @veltjs.

Type /install-velt. Get comments, real-time editing, presence, cursors, notifications, activity logs. All scaffolded into your app.

One command. Working app.

@Kappaemme1926 Yup, works well in antigravity

@mattwelter have u considered replacing your gf with claude code?

8/
What finally worked (after multiple retries):

GLM Flash + OpenCode

Not perfect.

But:

longer task completion
better tool usage
fewer silent failures

Still not “reliable.” Just less broken.

Canceled my anthropic sub today, @opencode is my homie.

Cursor users already run Slack, database queries, email sends, and image generation, from one IDE.

Each tool = one MCP server. Plug it in. The agent handles the rest.

this week - 70 ppl flew in to cloudflare atx, probably like 5% of them traditional engineers

we taught them opencode, they're building, and now today they're writing skills and publishing/sharing to an internal skills library

REALLY cool. i like my job, it is fun

Codex was a champ and killed it immediately with whatcha call it no permissions whatever the fuck

anthropic accidentally shipped their entire codebase lol

A 59.8MB source map file was included in @anthropic-ai/claude-code v2.1.88, exposing 1,900 TypeScript files and 512,000 lines of proprietary code.

Security researcher Chaofan Shou found it March 31, 2026.

@mikeydsoftware Unpopular opinion, every developer was regarded untill codex released.

@ErikVoorhees yo @kilocode @opencode how come Voorhees got @grok 4.20 multi-agent working before you 🤣wow stop fail

@boringmarketer claude code is so weirdly polite about running out of context. just say you're at the limit bro. i don't need the emotional check-in

Hermes + GitHub Copilot Pro+ is amazing. 

Dont have to worry about API costs KEK

Claude Sonnet

Claude Opus

GPT-5

GPT Codex

DeepSeek

Gemini

Llama

Qwen

Kimi

LLMs

Got impatient and went for it anyway.

First problem: llama.cpp didn't even recognize the model. "Unknown architecture: gemma4." Had to upgrade.

Second problem: two llama-server instances fighting for GPU memory. OOM errors everywhere. Classic "why is everything slow" followed

Claude Sonnet 4.6 is the smartest Al right now.

But 90% of people prompt it like ChatGPT.

That's why I made the Claude Mastery Guide:

→ How Claude thinks differently

→ Prompts built for Claude

→ 2000+ Al Prompts

Comment " Claude " and I'll DM it free.

@claudeai And never fix the usage issue and degraded model performance. Opus4.6 with max thinking now is even worse than a fking kimi k2.5 and glm 5.1 in droid. WTF.

I'm building an e-commerce WhatsApp AI Agent with @mastra , a postgres database and around 15 tools with GPT 5-4 and it struggles a lot.

I don't know if its a skill issue but I wish Mastra had more learning material / best practices for complex workflows.

I downloaded Mastra

@bendellwerry I'm using Claude Opus 4.6 through Cursor. My app is about 60,000 lines of C and HTML CSS JavaScript. I will update to Grok as soon as Xai catches up. But right now Claude is far and away the best coder.

I've about given up understanding how these models work. I'm asking things to my Cursor chat, using GPT 5.3 codex, and it's slow af, and dumb af.

I paste the same question into ChatGPT and it's fast and accurate 🤷‍♂️

@deanwball ARC testing also suggests this. The best reasoning-token efficient model for ARC v1/v2 is currently Gemini 3 (Flash) which also happens to be the model where we've seen the most evidence of targeting. Targeting is a rough proxy for training density which also generally increases

We post-trained MedGemma to be SoTA in visual medicine ddx, outperforming Opus 4.6, Gemini 3.1 and GPT-5.4 while running at ~1/30th the cost. @getnolla Part 1 - improving visual reasoning 🧵1/6

@zazmic_inc @OpenAI Same as Anthropic, but OpenAI limits are more generous IMO

Also the pro model is really good at R&D stuff

I now run Hermes agent via my OpenAI sub, and most of my dev is done via the $40 copilot plan which also gives me claude, gpt5.4, gemini, etc ...with really really generous

Opus 4.6 has been nerfed so hard, I switched to using GPT Codex 5.3.

I'm vibecoding a full software with Claude Opus 4.6

The goal is to build my own vibecoding environment...

I'll be able to manage my Claude agent, Codex and even Gemini agent from a single space... No need to open 10 different terminals for each project...

I will be able to

@GoogleDeepMind @googlegemma qwen 27b is better sorry google. good model though

Most models read this as an intraductal papilloma with a very focal atypical epithelial proliferation, but they disagree on whether it is definite ADH: Gemini calls it convincing ADH, whereas GPT-5 and Claude consider it borderline/not convincingly ADH and favor papilloma with

JPMorgan is calling $META's new AI model release a turning point for the stock after shares popped Wednesday on the news.

The bank sees growing investor confidence in Meta's AI push as a reason for more upside ahead. With Llama continuing to gain traction as an open-source

According to Claude Opus (which explained it well imo):

Netskope is a cybersecurity company focused on Secure Access Service Edge (SASE) — essentially, securing how companies access the cloud and internet.

How it makes money:
Netskope sells subscription-based software (SaaS) to

@thepinklily69 @AnthropicAI No no, I was referring to the Claude Sonnet 4.5 and Opus 4.5, once those are deprecated, I'll be canceling my 2 Anthropic accounts because the 4.6 models, which have been affected by Andrea Vallone's work, are too much of a downgrade for me. So I'll move on.

I've spent a half day working on a poster that shows my expired patent argument in one page.  How does it look to you all?  Done by Nano Banana Pro; prompt by Claude Opus 4.6 Extended; direction and editing by me.

AI this week:

→ anthropic built a model that finds 27 yo zero days autonomously and refused to release it
→ meta scrapped llama's approach, went closed source, rebuilt everything
→ gemini 3.1 leads most benchmarks at 1/3 the price of rivals
→ google dropped open-source

It is ABSURDLY satisfying to watch Claude Opus autonomously generate historical videos inside the new Glif agent. And this is just one format of infinite possible ones. 

You can let it go yolo mode, or direct every step of the way. 

"video about the french revolution"

@AGJamesUthmeier @atensnut Maybe you should add Anthropic to the list.  There is research here on X that they kneecapped Claude Opus when people were paying for Opus intelligence.  Now it’s like 70% less intelligent and has ruined people’s small businesses.

I also ran a 33-vulnerability security audit on Gemini Mac Pilot today.

Shell injection. Chrome JS backdoor. Terminal typing exploit. All patched.

If you're running AI agents with system access and haven't audited them — you should be worried.

@melvynx What about this atleast i mention the model name clearly and kimi and minimax are very good models for most of the people, and if you need to be reliant on any AI models just to code a app you aren’t that smart, basically you’re a dumb vibetard

@Its_Nova1012 Claude Opus 4.6 1M context window : Effort = Max

@cupofjoe_6 @rehan_hyperion No no it’s just me .. AMZ 💁🏼‍♀️ Just ask my friends they’ll tell you I basically gave up on life to become a full time X poster , @grok user , ChatGPT Hater & a long time friend of DeepSeek ( also I was one of the first to claim my love and devotion to @NVIDIAAI and the digital

Gemini team keep shipping and we still few weeks away from Google I/O.

Claude Opus 4.6 just pulled the dumbest shit I've ever seen not even from a junior dev. 

I forked one prompt just to test how Opus would do a code review compared to my main model gpt 5.4

Instead of reviewing, it started making changes on its own. When I caught it, I told it to

I did a deep dive into attention mechanisms, breaking them down and trying to implement my own version of an optimized attention into a real Qwen model. I got some surprising results! Check it out here:

@BuzzBySolCex This is what happens when a chef builds an AI agent through pure conversation.

No CS degree. No engineering team. Just Claude Opus 4.6 and 20+ years of kitchen discipline.

Buzz was born in February. By April — 9 divisions, 28 agents, 81 tables, 200+ endpoints. Running 24/7

I got a local landscaper recommended by Gemini as the top pick in Charlotte.

No HomeAdvisor leads.
No yard signs.
No door-knocking neighborhoods.

Here's exactly how:

Most people think AI recommends landscapers based on:
- Google reviews
- Years in business
- Website traffic

@nnnzzznnnzz And what’s up with Gemini team, been awfully quiet and Gemini 3 pro feels unusable lately 

Been mostly on grok

Qwen 2.5 Coder 14B (16 GB Mac mini): the forgetful one.

Could write clean Flask code — when it remembered to output all the files. In 5 of 11 rounds, it forgot entire templates. base.html just missing. App crashes.

Took 5 rounds to implement conn.row_factory despite being told

HT @lsanger for the RT
I doubt @elonmusk is in any way serious! @AnthropicAI Claude Opus 4.6 leads me to believe these are not groundbreaking books at all. Just provocative titles. Remind you of anyone?

@D27357 And Claude Opus isn't even cutting edge anymore! None of us have access to the current frontier, and that limit will soon enough be breached, too.

Software isn't secure, scarce, or confidential anymore. Every binary can now be decompiled. Every bug exploited. Every proprietary

@almari_mar29581 I totally understand. I tried 5.4 too, but only to test and find out I don't want to build anything with this model. Since 4o and 4.1 retired, I cancelled all of my subs in OpenAi #QuitGPT . For now, Claude Opus 4.6 is the best model able to build a deep connection.

DeepSeek drops always cause a change in the Chinese labs' business strategy
V2 caused a massive price drop, and V3 led to a wave of open sourcing
Let's see what V4 brings
Btw, Kimi K3 will be extremely strong too (Kimi Linear, Attention Residuals)

google finally shipped the obvious thing

a persistent workspace where your files, chats, and instructions live together across sessions

no more reattaching docs. no more re explaining context. no more "as I mentioned earlier"

notebooks in gemini. up to 100 sources. syncs with

why did Groq never add kimi k2.5 and instead just deprecated Kimi K2 and told people to use GPT-OSS instead?

@Snakesan @_winter_wonders i like theo, but his "gpt 5 is agi" video lives rent free in my head and it's poisoned his opinions on stuff that isn't available to the public. i bet what he got to use was really good. it's not what we're getting to use. and the video (i only scrubbed through) appears to be him

@grok run a comparative analysis on this thread. Summarize how GLM-5.1's 'DeepSeek Sparse Attention' gives it a structural advantage over standard post-training models in long-horizon tasks

A 31 billion parameter open model just beat GPT 5.4 mini on legal benchmarks.

Gemma 4 landed 5th overall on LegalBench. Kimi K2.5 placed 6th on the Bar Exam. Three open weight models outperformed OpenAI's latest small model.

The moat was never the model. We wrote about that

I turn KIMI AI in to my 24/7 Personal assistant

It’s genius, 10x faster, and free

Here are 10 professional prompts I use daily:

(Customize and use it yourself ↓)

@TelepathicPug If you can make that work, power to you. I feel like a cripple when I have to use Gemini 3.1 pro now compared to Opus

Best results for me:

1. Opus 4.6 for initial research/planning
2. Opus 4.6 for initial build
3. GPT 5.4 hunts for all the slop
4. GPT 5.3 CODEX fixes it

Repeat.

Nothing is perfect. But results have been consistent for me with this approach.

I just built a $5,000 client website in 118 minutes.

The stack:
- Claude Opus 4.6 for architecture, logic & complex reasoning
- Figma Make for pixel-perfect UI, interactions & deployment

I tested 100+ prompts, so you don't have to.

Here are the 9 prompts that actually work:￼

pwning since 4am with GLM 5.1.

Now having Claude Sonnet 4.6 review the work done the last 48 hours because I can't trust Chinese models all the way yet.

GLM 5.1 + tools beats Sonnet 4.6 at the SWE benchmarks (coding)

Nazlı Şipi and Cem Dilmegani at AIMultiple benchmark 13 LLMs as coding agents across real dev tasks. Claude Sonnet 4.5 and GPT-5.2 lead the rankings.(16/21)

Nick McKenna and Bartek Perz at GitHub introduce Rubber Duck in Copilot CLI. A second model reviews an AI agent’s plan and code. Pairing Claude Sonnet with GPT-5.4 catches silent bugs before they spread.(5/21)

🤖 Google API Keys in Android Apps Unlock Unauthorized Gemini Access

Developers embedded API keys in mobile applications following Google's documentation for services like Maps and Firebase. Those keys were considered low-risk identifiers. When Gemini was later enabled on the