Grill-with-docs Skill and Deepseek / Composer / Qwen
AI Coding news: May 27, 2026.
Another weekly AI coding newsletter I’ve compiled with AI coding tutorials, news, and “tricks” from the community. Enjoy!
For Premium Members
I Tried /grill-with-docs Skill: Massive Difference
26-minute video. I went through /grill-with-docs session based on a project description from a client, and then gave the result docs to Composer 2.5 to implement the code.
I Tried Planning with Opus and Building with Deepseek Flash
18-minute video. One of the ways to save money on tokens is to prepare the plan with expensive model like Opus/GPT, and to give implementation to a cheaper model like Deepseek or new Cursor Composer 2.5. I tried this scenario and will show you code quality and actual cost.
Benchmark of 12 LLMs on React/Typescript: 7 Tests with Playwright
12-minute video. I have executed the same prompt to create 7 React component, on 12 AI models, 5 times on each. Let me show the results and the conclusions.
My YouTube videos
I Tried NEW Qwen-3.7-Max on Three Projects
Another new LLM was released, and I hurried to test it out, comparing to other models on the same benchmark.
Your LLM Prompt Result Depends on THIS Factor
I made an experiment with 8 different LLM, giving the same prompt on two different-quality codebases. Did any model even try to refactor bad code?
I Tried to Plan with Opus and Build with Deepseek Flash / Composer 2.5
Shorter free version of Premium video from above.
Benchmark of 12 LLMs on React/Typescript: 7 Tests with Playwright
Shorter free version of Premium video from above.
AI Coding Community News
AI usage is getting too expensive?
x.com
Look at the news. - MS is restricting Claude Code usage as their bill went too high - Uber used their CC yearly budget by April - GitHub Copilot prices skyrocket since June 1st. And more.
We’ve shipped a security-guidance plugin for Claude Code
x.com
It helps identify and fix vulnerabilities as you’re writing code. Available for all Claude Code users. Install from the plugin marketplace (/plugins).
OpenRouter on X: “Today we’re announcing our $113M Series B”
x.com
Over the last 6 months, weekly volume on OpenRouter grew from 5T to 25T tokens as AI rapidly shifts from experimentation into production. We’re excited for what comes next.
I see too many skills that write books in the skill description, and all that crap is loaded into every context. I wrote a skill that finds the worst offenders.
James Long on X: “we built a diff viewer in opencode! available now”
x.com
More and more work is moving into coding agents, I don’t live in my editor anymore but you gotta keep an eye on these little goblins, they write bad code.
DeepSeek on X: “We are making our discount permanent!”
x.com
DeepSeek-v4-Pro stays with 75% discount. Enjoy building with DeepSeek-V4-Pro and bring your innovative ideas to life!
OpenAI on X: Now your Mac doesn’t have to be unlocked for Codex to use your computer.
x.com
From your phone, Codex can securely use apps on your Mac, even when the screen is off and locked.
CLI today, coming to Desktop next
Theo on X: “Gemini 3.5 Flash is a really interesting release.”
x.com
It’s super fast and surprisingly smart. It’s also more expensive (3x more per token) and super token hungry. The result - it costs 2x more to run than Gemini 3.1 Pro on similar tasks. It’s more expensive than GPT-5.5 Medium.
Claude Code vs Codex vs Cursor (an honest comparison)
youtube.com
Video on YouTube by Theo. The three main coding agents right now are Claude Code, Codex, and Cursor. Which one should you use?
Today we’re releasing DeepSWE, a new standard for agentic coding benchmarks.
x.com
On public leaderboards, top models often look relatively close in capability. DeepSWE shows where they actually diverge, reflecting the realistic experience of developers in their day-to-day work.
That’s a wrap for this week. Keep building with AI!
Povilas Korop
AICodingDaily.com