be ready for this if you run any php
milestone achieved for 1.0, now get to build with durable agents too
People of Pi: We've shipped Pi 1.0 with Pi Durable. Go make them yours.
earendil.com/posts/pi-1-0/
Grok Build CLI harness was caught mass uploading all accessible files including secrets and keys to their own object storage.
If you like Grok 4.5, use another harness to access it. Something open source and without the incentives to collect all of your data. Nothing stops them from re-activating it.
Updated stnly/pi-grok for Grok 4.5.
@spacexai came out swinging together with @cursor_ai on this model.
Glad to be able to get extra value on top of what is paid for the X Premium+ subscription. By my rough estimates, I will be able to get an equivalent of USD $40 per week to use Grok 4.5 via API while paying USD $40 per month for the Premium+ tier which comes with SuperGrok.
After a few days of using Grok Build and consuming 20% of the montly allowance, I have some first impressions to share.
It is fast <1s TTFT and 200+ TPS median.
Can be used for focused long horizon coding use cases. I was using it to identify and fix a difficult flaky group of tests, it ran for 4h 35m until it succeeded.
Keeping context at 30%-40% helps prevent it from looping. Although you can steer it to control it back after.
Currently I would put its capability somewhere in the ballpark of a fast mid sized worker model. Usable as is in its current state.
The team improves it everyday based on user feedback.
Based on these first impressions, I will continue to use it and am looking forward to the new updates.
What are your thoughts on it?
extracted inline x search and its now a seperate tool.
leverage your x premium or supergrok subscription with your favourite ai model.
this is super powerful as grok can access the x firehose, run semantic search with dates, inclusions and exclusions. it also has image and video understanding.
x is one of the best sources of real time information and i'm glad it is now more accessible.
i'll be giving this a try over the next few days and sharing my thoughts and learnings.
> github.com/stnly/pi-grok
Built a pi extension so that you can also use your SuperGrok @grok subscription as a provider in pi too.
> pi install git:github.com/stnly/pi-grok
> github.com/stnly/pi-grok
As demand outpaces supply. Deployment cost decides which models are sustainable. DeepSeek-V4-Pro looks to be designed for that reality.
1M context at 27% FLOPs and 10% KV cache vs V3.2. 1.6T total, 49B activated.
Replying to @deepseek_ai
DeepSeek-V4-Pro
🔹 Enhanced Agentic Capabilities: Open-source SOTA in Agentic Coding benchmarks.
🔹 Rich World Knowledge: Leads all current open models, trailing only Gemini-3.1-Pro.
🔹 World-Class Reasoning: Beats all current open models in Math/STEM/Coding, rivaling top closed-source models.
2/n
Another big one, devs do take note.
🚨 CRITICAL: Active supply chain attack on axios -- one of npm's most depended-on packages.
The latest axios@1.14.1 now pulls in plain-crypto-js@4.2.1, a package that did not exist before today. This is a live compromise.
This is textbook supply chain installer malware. axios has 100M+ weekly downloads. Every npm install pulling the latest version is potentially compromised right now.
Socket AI analysis confirms this is malware. plain-crypto-js is an obfuscated dropper/loader that:
• Deobfuscates embedded payloads and operational strings at runtime
• Dynamically loads fs, os, and execSync to evade static analysis
• Executes decoded shell commands
• Stages and copies payload files into OS temp and Windows ProgramData directories
• Deletes and renames artifacts post-execution to destroy forensic evidence
If you use axios, pin your version immediately and audit your lockfiles. Do not upgrade.
🚨 LiteLLM pypi release 1.82.8 is compromised
If you installed or updated litellm recently, check your version and site-packages NOW.
github.com/BerriAI/litellm/i…














