Building & testing AI products. Sharing what works, what breaks, and what I learn.
Earth
Joined January 2026
- Tweets12.8K
- Following951
- Followers2.7K
- Likes12.3K
Buildwahab retweeted
I am genuinely surprised DeepSeek V4.1 Flash (FREE) inside OpenCode built this 🤯
A fully interactive 3D mechanical spider.
8 articulated legs, 7 movement modes, exploded views & internal inspection.
All in single HTML file.
This is wild 🕷
Demo + source code below 👇
Buildwahab retweeted
Built a fully interactive 3D Mechanical Crab with Claude Opus 5.5 🦀
Rotate it. Disassemble it. Explore its internals. Watch it move.
From a simple prompt to a live 3D experience.
AI is getting crazy
Live demo + source code below 👇
Buildwahab retweeted
🚨 Gemini 4 benchmark leaks 👀
> Allegedly ahead of GPT-5.6 Sol + Fable 5
> Reportedly much stronger at coding
> Massive 1.5M-token context
Built for long-running agentic workflows
>Significantly larger frontier model
If this is real… Google might be cooking. 👀
Buildwahab retweeted
SpaceX has officially acquired Cursor.
So… what are we renaming it?
CursorX?
StarCursor?
SpaceCursor?
I’m voting for CursorX.
Buildwahab retweeted
GLM-5.3 is LIVE🔥
Built on the same 743B base as GLM-5.2, but pushing performance to a whole new level.
- Terminal Bench 3.0: 28.3
- DeepSWE: 66.9
- Agents’ Last Exam: 28.5
- GDPVal-AA: 1769
And it’s reportedly matching models several times its size. 👀
Introducing GLM-5.3: Built to Code. Ready for Cyber Defense.
- Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model
- A major leap in cybersecurity, setting a new standard among open models
Tech Blog: z.ai/blog/glm-5.3
Buildwahab retweeted
The next wave of frontier models is already looking pretty stacked:
- Grok 4.7
- GPT-6
- Fable 5.1 / Opus 6
- Gemini 3.5 Pro
- Kimi K4
The crazy part is how fast the leaderboard keeps changing. Nobody stays on top for very long anymore.
Which one are you most excited to see?
Buildwahab retweeted
Gemini 3.7 Flash could drop today 👀
And the price is wild reportedly just $0.75/M input and $3.75/M output, roughly half the cost of 3.6 Flash.
Google might be cooking. 🔥
Buildwahab retweeted
🚨 GPT-6 / Astra Leak 👀
OpenAI's rumored next major model, Astra, is sounding absolutely insane.
- Built for long-running multi-agent tasks
- Reportedly solved 10 major math & CS problems
- Could launch as GPT-6, GPT-5.7, or under the Astra brand
If even half of these leaks are accurate... we're about to witness another huge AI leap.
Bullish on what's coming.
Buildwahab retweeted
Just spotted Gemini 3.5 Pro on Arena Battles 👀
Bullish. Gonna put it to the test today
Buildwahab retweeted
Elon says Grok 4.6 is coming around August 7 a 1.5T parameter model with significantly improved SFT & RL.
Then just a few weeks later: Grok 4.7 at 2.1T.
xAI is accelerating fast.
Buildwahab retweeted
Grok is moving like Flash. ⚡
4.6 in 2 weeks.
4.7 in 4 weeks.
Now I'm curious will Grok 4.7 be able to compete with Kimi K3 and Fable 5?
Buildwahab retweeted
Does Gemini 3.5 Pro really beat Claude Fable 5?
That's what the latest leaks are claiming.
- Reportedly outperforming Claude Fable 5 and GPT-5.6 Sol on multiple internal benchmarks
- New Deep Think reasoning mode
- Up to a 2M token context window
- Major improvements in coding, planning, and AI agents
- Better frontend generation and higher-quality SVG output
- Internal codename: Cappuccino
- Still undergoing testing with select partners
If these leaks turn out to be accurate, Google could be making a serious comeback in the AI race.
Do you think Gemini 3.5 Pro will live up to the hype?
Buildwahab retweeted
As someone who's been building with Grok recently, this is great news.
Open-sourcing Grok Build gives developers the chance to understand how it works and help make it even better.
Looking forward to seeing what the community builds.
We've open-sourced Grok Build and have reset usage limits for all users.
Open sourcing Grok Build allows anyone to support making a reliable and robust harness. Check out our code, including the Git repo for the Grok Build CLI.
x.ai/open-source
Buildwahab retweeted
Yesterday: 30% of my weekly SuperGrok limit gone.
Today: Back to 0%.
Looks like xAI surprised SuperGrok users with an early usage reset.
Time to build.
Buildwahab retweeted
About 2 hours left until my weekly Grok limit resets. 👀
Planning to build something exciting today.
Hopefully I can finish the project before I hit the limit again. 😅
For those who use Grok regularly, how do you make your limits last longer?
Any tips or workflow tricks?
Buildwahab retweeted
Gemini 3.5 Pro is shaping up to be one of the most anticipated AI releases of the year.
Recent leaks suggest:
- Launch has reportedly been delayed to further improve the model
- New Deep Thinking reasoning mode
- Up to a 2M token context window
- Stronger coding, tool use, and agentic workflows
- Broader testing appears to be underway
If the rumored benchmark gains hold up in real-world use, the AI race is about to get even more interesting.
Are you looking forward to Gemini 3.5 Pro?
Buildwahab retweeted
GPT-5.6 is here.
Grok 4.5 is here.
Now all eyes are on Gemini 3.5 Pro.
Rumors suggest:
- Better performance than Claude Fable 5 and GPT-5.6 in internal testing
- Major zero-shot improvements
- Currently in private validation
- Public launch reportedly targeted for July 17
If the leaks are accurate, the AI race is about to get even more interesting.
Are you looking forward to Gemini 3.5 Pro?
Buildwahab retweeted
Grok 4.5 Test: It really beats Opus in the trenches 👀
> Stronger at messy real-world codebases without constant hand-holding
> Much faster iteration loops on actual features
> Far less "I can't help with that" moments
> More token-efficient for long debugging threads
> Built for devs who ship, not just benchmark chasers
Would you switch from Opus to Grok 4.5?
