Skip to main content
Back

OpenAI Codex Mobile Control , xAI Grok Build CLI , X Feed Algorithm

Musk open-sources X ranking algo, xAI drops Grok Build CLI, Nous Research cuts LLM pretraining 2-3x, Alibaba shrinks Qwen 80B to 23B

May 15, 2026 6 min read

Hey,

The coding agent wars just went fully mobile and terminal-native in the same 24 hours. OpenAI put Codex in your pocket. xAI dropped Grok Build in your terminal. Alibaba is shrinking models so they run anywhere.

The pattern is clear: agents need to be always on, always reachable. Not just smart, but ambient.

We're watching the interface layer of AI development get rebuilt in real time. Your desk is optional now.
Together with · partner_image
Author photo
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
Summary

Read time: 6 min 50 sec

Top Repo

▸ Elon Musk open-sources X's feed ranking algorithm on GitHub

Fin

▸ Join Fin Labs in NY from June 2-4

Top News

▸ OpenAI brings Codex agent control to ChatGPT mobile on iOS and Android

AugmentCode

▸ Engineering Teams Are Feeling Two Things at Once

Top News

▸ xAI drops Grok Build, an agentic coding CLI for building apps and automating workflows

Signals

1. Nous Research releases training trick that cuts LLM pretraining time 2–3x at no extra cost
2. Presented by Tiger Data: Your agent makes decisions on stale data. Your users pay for it. TimescaleDB extends Postgres for live analytics at scale.
3. Made With ML teaches 40K developers to ship production-grade ML apps
4. Alibaba shrinks Qwen's 80B MoE model to 23B using pruning and distillation tricks
5. Unsloth ships Qwen3.5-9B with speculative decoding for up to 2x faster generation
6. GitHub opens waitlist for standalone Copilot app technical preview
Top Repo
Elon Musk open-sources X's feed ranking algorithm on GitHub
25,495 Likes
Elon Musk open-sources X's feed ranking algorithm

X just open-sourced the full code behind your For You feed. Not a blog post. Not a diagram. The actual production code, live on GitHub at xai-org/x-algorithm under an Apache 2.0 license, meaning you can use and modify it commercially.

Here is what the system actually does. Every time you open X, it pulls from 500 million daily posts, narrows them to roughly 1,500 candidates, and ranks them in under 200 milliseconds using a Grok-powered AI model.

The stack worth knowing:

  • Written in Rust (62.9%) and Python (37.1%)
  • Four components: Home Mixer (orchestration), Thunder (post storage), Phoenix (Grok-based ranking), and Candidate Pipeline
  • A pre-trained mini model ships with the repo, so you can run inference without training anything yourself
  • Code updates promised roughly every four weeks

What this opens up: you can study how a planet-scale recommendation system is actually built, borrow the pipeline patterns for your own projects, and finally stop guessing why your feed looks the way it does.

Read more
forward →
Presented by Fin
Join Fin Labs in NY from June 2-4

Fin Labs is coming to New York! For three days only, join a unique learning space in the heart of the city to explore transforming customer experience with AI.

Hear from leaders at Fin, Clay, and more on scaling AI for CX and see how they're transforming their organizations.

Standout sessions:

See the schedule
partner with us →
Top News
OpenAI brings Codex agent control to ChatGPT mobile on iOS and Android
21,477 Likes
OpenAI brings Codex agent control to ChatGPT mobil

Codex just landed in the ChatGPT mobile app, and it solves a real annoyance. You kick off a long coding task, step away from your desk, and then Codex hits a decision point and just... waits. Now your phone becomes the remote control.

Here is what you can actually do from mobile:

  • Review live outputs, diffs, terminal logs, and test results in real time
  • Approve or redirect Codex when it needs a decision to keep going
  • Start brand new tasks from scratch, right from your phone
  • Switch models or jump between threads across all your active work

Your files, credentials, and permissions never leave your machine. The phone is just the interface. Codex keeps running on your Mac, devbox, or remote environment.

To get started on Mac: open the Codex app, find Set up Codex mobile in the sidebar, scan the QR code with ChatGPT on your phone, and you are paired. Over 4 million people use Codex every week, and now none of them need to be at a desk to stay unblocked.

Read more
forward →
Presented by AugmentCode
Engineering Teams Are Feeling Two Things at Once

“Optimistic, excited, threatened.” “Insecurity, mistrust, hope.” We asked 219 engineering leaders how AI-native engineering feel, and why most org systems still aren’t built for this new reality.

Get the full State of AI-Native Engineering 2026 Report.

Read the report
partner with us →
Top News
xAI drops Grok Build, an agentic coding CLI for building apps and automating workflows
9,393 Likes
xAI drops Grok Build, an agentic coding CLI for bu

xAI just dropped Grok Build, a coding agent that lives in your terminal. Think of it like a smart assistant that does not just suggest code, it actually takes action: reads your repo, proposes a plan, edits files, runs shell commands, and checks its own work.

Here is what makes it stand out:

  • Parallel subagents: it splits big tasks across multiple agents running at the same time, so complex jobs finish faster
  • Plan Mode: you review and approve the full plan before anything executes, no surprises
  • 2 million token context: it can hold your entire large codebase in memory at once
  • Headless mode lets you run agents inside scripts and automations

Install it with a single curl command, sign in, and you are in. No API key setup, no config files. To try it:

curl -fsSL https://x.ai/cli/install.sh | bash

The catch: it is locked behind SuperGrok Heavy at $300/month. It is an early beta, so rough edges exist. Type /feedback inside the CLI to send issues straight to the team.

Read more
forward →
Signals
1
Nous Research releases training trick that cuts LLM pretraining time 2–3x at no extra cost
4,553 Likes
2
Your agent makes decisions on stale data. Your users pay for it. TimescaleDB extends Postgres for live analytics at scale.
Presented by Tiger Data
3
Made With ML teaches 40K developers to ship production-grade ML apps
47,608 Stars
4
Alibaba shrinks Qwen's 80B MoE model to 23B using pruning and distillation tricks
548 Likes
5
Unsloth ships Qwen3.5-9B with speculative decoding for up to 2x faster generation
2,217 Downloads
6
GitHub opens waitlist for standalone Copilot app technical preview
4,463 Likes
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 200,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.

Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
Looking to promote your company, product, service, or event to 250,000+ AI developers?
Work with us