Composer 2.5, Claude Prices, and AI "Tricks"
AI Coding news: May 20, 2026.
Another weekly AI coding newsletter I’ve compiled with AI coding tutorials, news, and “tricks” from the community. Enjoy!
For Premium Members
Codex App: 8 Features You May Not Know
11-minute video. Codex App is really convenient to work with multiple projects at the same time. I've found quite a few "hidden tricks" that are not visible right away. Let me show you.
My YouTube videos
I Tested NEW Composer 2.5. Wow. (Updated LLM Benchmark)
I have benchmarked the new Cursor Composer 2.5 model with the same projects/prompts I did earlier this month, let me show the results.
AI Coding “Trick”: Ask LLM Why It Made Certain Decisions
I saw a tweet by Thariq from Claude Code which appeared to be incredibly powerful to analyze the prompt result. Let me show you, in Claude Code and Codex.
I Realized Why Western LLMs Beat Chinese Models: My Example
My new benchmark shows Western/Chinese models in action, and important detail that I found repeating.
I Tried NEW Clawpatch for Codex: “Optimized” Code Review
I saw this tweet by Peter, author of Openclaw, and decided to try his Clawpatch.
AI Coding Community News
The credit covers usage of:
Claude Agent SDK
claude -p
Claude Code GitHub Actions
Third-party apps built on the Agent SDK
Matt Pocock on X: “Anthropic has given us a “dedicated monthly credit”. What does it ACTUALLY mean?
x.com
It slashes AFK usage limits of Claude Code by ~5-20X. Here’s how it affects you.
Claude Code weekly limits are increasing 50%, now through July 13.
x.com
Live now for all Pro, Max, Team, and seat-based Enterprise users.
The Claude Code Setup Behind Shopify’s 23,000 Engineers (Exact Config You Can Copy)
x.com
Long-form article on X by @zodchiii
ClaudeDevs on X: “What are best practices for running Claude Code at scale?”
x.com
New blog post on what we’ve learned from teams running it across multi-million-line monorepos, decades-old legacy systems, and distributed microservices.
Cursor on X: “Introducing Composer 2.5, our most powerful model yet.”
x.com
It’s more intelligent, better at sustained work on long-running tasks, and more reliable at following complex instructions. For the next week, we’re doubling the included usage of the model.
Send your system prompt before the user prompt. Claude writes it to the cache, but skips generating any output. When the real user request lands, it’ll hit a warm cache.
OpenAI on X: “Now in preview: Codex in the ChatGPT mobile app.”
x.com
Start new work, review outputs, steer execution, and approve next steps, all from the ChatGPT mobile app. Codex will keep running on your laptop, Mac mini, or devbox.
Povilas on X: “MiMo-v2.5-Pro in OpenCode is often auto-switching into Plan mode”
x.com
It suddenly auto-switched into Plan mode (in Opencode), so when I got back to computer in 10 minutes, planning to review the FINISHED task, it was like “here’s the plan, should I go ahead?” For a pretty trivial task.
That’s a wrap for this week. Keep building with AI!
Povilas Korop
AICodingDaily.com
23,000 engineers at Shopify on Claude Code is the enterprise adoption number that reframes everything else in this newsletter. At that scale, the 50% usage limit increase isn't a product perk — it's a supply-chain decision that changes Shopify's engineering cost structure. The 5-20X time-to-first-token reduction from prompt caching moves from 'nice to have' to 'required' as teams hit token ceilings. At theaifounder.substack.com I write about the builder side: the teams getting most from these tools aren't just using them faster, they're restructuring how work gets decomposed. What's your sense of whether the Composer 2.5 usage doubling reflects genuine productivity lift or engagement from novelty?