Just left the Cafe for a bit, kudos to @benln for a really awesome event! Truly a co-working and co-building session 💻

At this point, I don't think I can type anywhere without autocomplete

GPT-5-Codex is now available in Cursor. Let us know your thoughts!

We've trained a new Tab model that is now the default in Cursor. This model makes 21% fewer suggestions than the previous model while having a 28% higher accept rate for the suggestions it makes. Learn more about how we improved Tab with online RL.

MoE layers can be really slow. When training our coding models @cursor_ai, they ate up 27–53% of training time. So we completely rebuilt it at the kernel level and transitioned to MXFP8. The result: 3.5x faster MoE layer and 1.5x end-to-end training speedup. We believe our

cookbook for @cursor_ai cli is added with examples of - auto fixing ci failures - updating docs - secrets scanner - automatic i18n

Cursor CLI now includes MCPs, Review Mode, /compress, @-files, and other UX improvements.

This is truly AI week. Welcome Cursor CLI:

To help you along with the GPT-5 release (and free initial usage in @cursor_ai!), check out this model prompting guide from @ericzakariasson, Anoop Kotha and Julian Lee https://t.co/DQZVs395ua

GPT-5 is really strong, it's one of the few models I don't need to switch off for certain tasks. But I'm really just happy to make it through a week of live demos without spilling the beans

Cursor 1.4 is out with a significantly more capable agent. It’s now much better at challenging and long-running tasks, especially in large codebases. We’ve also given the agent better tools, made token usage more efficient, and improved code editing accuracy.

Cursor 1.3 is out! You can now collaborate with Agent in your terminal, clearly see context window usage, and make faster edits.

In the past month, Cursor found 1M+ bugs in human-written PRs. Over half were real logic issues that were fixed before merging. Today, we're releasing the system that spotted these bugs. It's already become a required pre-merge check for many teams.

Ever wonder if your agent’s actually getting it right over a whole convo, not just one step? New Session-Level Evals in Arize AX let you do exactly that by measuring: 🌀 Coherence across the session 🧩 Context retention across turns 🎯 Whether users actually reach their goals

In case you missed some big news from Arize Observe 2025: Phoenix Cloud just leveled up with Spaces & Access Management ✨You can now create multiple, tailored Phoenix Spaces for your team and projects 🔑 Easily manage user permissions in each space 👥 Zero-hassle team

Today's the day!🎉 Arize Observe just kicked off, and it's bringing a whole set of new product announcements. From Agent-powered trace debugging to new Prompt Learning techniques, we've got it all! Announcements in the thread below 🧵 👇

🌟 Observe 2025 kicked off with a packed keynote We just dropped a stack of new features across Phoenix Here’s what’s new 👇

How do you evaluate a whole crew of AI agents, not just a single one? 🤔 With @JohnGilhuly from @ArizeAI, we created an example demonstrating how to build a multi-agent system using CrewAI, develop a reference dataset for ideal task sequences and use Vertex AI's Gen AI Eval and

New visualizations to track your experiment evals and latency in @ArizePhoenix 👀📊 We've made it easy to clearly see how your experiments evolve over time. This has already saved me time I would've spent on manual digging. I can clearly see how performance shifts & more

Okay, all you @cursor_ai fans out here Imagine Cursor, but debugging across all instantiations of observability (traces/sessions), evals, and iterations It's going to be a good year for @arizeai