@buildsolo

Read the actual code too. The most underrated finding: "not_dwelled" is a separate negative weight. People scrolling past your post actively hurts you — it's not just neutral anymore. Also dwell_time is continuous, not binary. The algorithm measures exactly how many seconds someone reads. Longer posts with substance literally score higher per-reader.

The latest 𝕏 algorithm is now on GitHub. So I used Grok Build to analyze it. The biggest takeaway: X is no longer a simple likes/replies system. It's a Grok-powered recommendation engine. 10 things creators need to know:

Just cloned this and read the scoring code. Key finding: "not_dwelled" is now a negative signal. If people scroll past your post without stopping — your score drops. Dwell time (how long they read) is a continuous weight, not binary. Translation: long posts with real substance > short motivational one-liners. The algorithm literally rewards depth.

Just cloned this and read the scoring code. Key finding: "not_dwelled" is now a negative signal. If people scroll past your post without stopping — your score drops. Dwell time (how long they read) is a continuous weight, not binary. Translation: long posts with real substance > short motivational one-liners. The algorithm literally rewards depth.

The latest 𝕏 algorithm has been published to GitHub github.com/xai-org/x-algo…

@JonBuildsHQ Yeah, similar concept but different execution. Claude Code is terminal-only — you type commands, it works in your repo. Kiro is a full IDE (VS Code fork) with Claude inside. You see file changes live, get specs/hooks system, and can deploy directly from the chat.

@buildsolo_x Awesome! It's slightly different from something like Claude Code, right?

Founders/Builders: What's the dev/productivity tool you cannot imagining living without right now?

If I may, I will explain using an example from my own work. Today I used it to: update a Cloudflare Worker, deploy it, rollback when the model broke, restructure project files — all from one chat window. No context switching. I describe what I need, it reads my codebase, writes the code, runs the commands. Feels like pair programming with someone who already knows your project.

@buildsolo_x Sounds interesting! How does it help you?

@JonBuildsHQ I think Musk would not be happy😅

@buildsolo_x For a a second I thought I had cracked the algorithm

What is going on? 😭

This works. I went the opposite route — I just tweet what I shipped that day. Today: tested Gemini 3.2 Flash in production, translations broke, rolled back in 20 min. That's the tweet. No batching needed when you're building daily. But for scaling content — your system makes total sense.

It used to take me an hour to write 5 tweets. Now I write 240 monthly tweets in 4-6 hours. Create a content system. It'll change your life.

@pankajkumar_dev Yeah exactly. I wanted to test 3.2 before the official release — the endpoint exists as "gemini-3-flash-preview" but the output quality isn't there yet. Incomplete translations, cut off mid-sentence. Rolled back to 3.1 Flash Lite in 20 min.

@buildsolo_x ohh u mean 3.2 u tested and now they rolled back, many ppl seen this btw.

Gemini 3.2 Flash leaks: fast and cheap seems to be the focus - Gemini 3.2 Flash looks focused on making AI much faster and cheaper without sacrificing too much quality - According to my sources, Google may rename it to Gemini 3.5 Flash - It may perform close to Gemini 3.1 Pro level while keeping very low latency with sub-200ms responses rumored for many queries - Pricing leaks point to around $0.25 input / $2 output per 1M tokens, though honestly that still feels too cheap to fully trust right now - Google is using stronger distillation and sparsity techniques to compress larger model capabilities into a lightweight version - Knowledge cutoff is said to be updated to January 2026 - Google also seems focused on grounding + search reliability to reduce hallucinations in real-world workflows - Expected around Google I/O, possibly 1-2 days before the keynote

@pankajkumar_dev 3.2, although not released, I connected my API to it, and the result was not good. I rolled back to my Gemini 3.1 Flash Lite. I tried.

@buildsolo_x 3.2/3.5 Flash not officially released yet.

For context: the 1.1 sec includes TLS handshake, Worker processing, Gemini inference, and streaming back. Warmup prefetch alone cut cold-start from 4s to 700ms. Next experiment: edge caching for repeated translations. Same text twice = instant, zero API call. What latency do you consider acceptable for a real-time translation tool?

Optimizing BlackRabbit for users in Asia right now. Problem: data centers are in the US. From Vietnam, every translation request travels 15,000 km round trip. Current result: 1.1 sec per translation. Honestly — not bad for that distance. Goal: get it down to 500-800ms. Working on warmup prefetch, adaptive timeouts, and smart routing. Geography is the hardest bug to fix. #vibecoding

@TTrimoreau I agree with you, my friend! Nowadays, creating a product has become very accessible. You just need to understand the tools a little bit and convey your idea to the AI as effectively as possible.

You don't need to know data structures and algorithms to build real products

@IndieGameJoe Wow, this is just a brilliant game!

This indie team is making a kart racing game where your vehicle is an office chair - Drift office chairs through hallways - Each chair & driver has unique perks - The longer you drift, the faster you go Would you play this? It's called Need for Seat

@bindureddy This is an interesting confrontation. These two giants are making very strong projects. I'm waiting for the release of Gemini 3.2 so I can install it in my SAAS.

Next week is going to be crucial Will Google pull ahead of OpenAI or vice versa? GPT 5.6 or Gemini 3.2 Who will win?

@mscode07 Congratulations on your first payment!

I made the first sale 🥳 And it's not even launched yet!!!!

@enjojoyy Be sure to try Vietnamese salted coffee.

I’m going to Vietnam to cowork with my teammates and friends for a month lowkey can’t believe this is my life but I’m really grateful for where I am and the things I can do

v2.0 is solid, but we're pushing further with a more premium AI layer. Can't say much yet, but if you're a Pro user — watch this space.

BlackRabbit v2.0 is live on Chrome Web Store. Now with Gemini 2.5 Flash-Lite. New UI. We shipped 4 things: 1. Speed & Quality — Gemini 2.5 Flash-Lite. Professional translation. Context, abbreviations, slang. 2. UI Update — Premium. Cleaner. Built for power users. 3. Focus Tools — Sounds and games for pleasant breaks. 4. Enhanced Animation — New animation system for better UX. BlackRabbit is built for those who work without borders. Download. Try. Tell us. #buildinpublic