Vyacheslav Legostin's Blog

The week of June 14-20, 2026 was not about one loud model launch. The louder story was that AI development is no longer just a contest over which model is smarter. It is becoming a contest over delivery, control, cost, standards, regulation, and the agent infrastructure around the models.

Viewed separately, the news looks like a normal feed: OpenAI published new research, Google and Microsoft backed a specification for discovering agent resources, GitHub added new Copilot surfaces and metrics, Vercel presented a production agent stack, Cursor was pulled into a huge acquisition story, and the Anthropic Fable 5 shock kept shaking the market. Put together, the pattern is clearer: agentic software is growing up, and the boring parts around it are starting to matter most.

Read more

Everyone "learns with ChatGPT" now — and a week later all that is left is a chat transcript you will never reopen. Chats answer questions; they do not teach. Real learning needs structure, practice, feedback and repetition. That bugged me, so I built a tool that turns "I want to understand X" into an actual course.

It is Learn (Almost) Anything — a free, open-source desktop app for macOS and Windows. You type a topic and an agent interviews you, researches the material, designs a curriculum, writes illustrated lessons, quizzes you, grades your homework and schedules flashcard reviews. And the key part: the engine is the very Claude Code or Codex CLI you already pay a subscription for.

Read more

Four days ago Anthropic launched Claude Fable 5 and Mythos 5, the most capable models the company had ever shipped. Today both are unavailable. On June 12, 2026, Anthropic disabled access to Fable 5 and Mythos 5 for all users in order to comply with a US government directive. In parallel — but for a completely different reason — Microsoft barred its own employees from using Fable 5.

This is an unusual case on several counts. It appears to be the first time a frontier commercial model deployed to hundreds of millions of people has been recalled by direct government order. And the story is already muddled: two separate "bans" keep getting collapsed into one. Let us untangle what actually happened and what follows from it.

Read more

Anthropic has launched two models at once: Claude Fable 5 and Claude Mythos 5. In short, Fable 5 is the public version of a new Mythos-class model, while Mythos 5 is almost the same model with broader capabilities enabled for trusted cybersecurity and biology teams.

The main story is not that the benchmarks went up again. That part is expected now. The more important shift is that Anthropic is trying to give regular users a model above the Opus class without opening the riskiest capabilities to everyone by default.

Read more

This is a Russian translation of Anthropic's article «Best practices for using Claude Opus 4.7 with Claude Code».

Read the original: claude.com

Read the Russian translation: legost.in/p/claude-opus-4-7-best-practices Read more

This is a Russian translation of Bassim Eledath's article «8 Levels of Agentic Engineering».

Read the original: bassimeledath.com

Read the Russian translation: legost.in/p/8-levels-of-agentic-engineering Read more

I got tired of waiting for ncdu to finish scanning. Built my own disk space analyzer — cull. The key difference: you can use it while it's still indexing — open it, immediately see the files, browse directories, while sizes are calculated in the background. Written in Go with the help of Claude Code.

More about features and installation: legost.in/en/utilities/cull Read more

I was born and raised in Almaty. I love this city. And for fifteen years now, I've been capturing it with my camera. Over this time, I've accumulated many shots that I consider successful.

This page is an attempt to convey my feelings for the city through photographs. It should by no means be taken as a comprehensive guide to the city or a catalog of attractions. There are a great number of wonderful places I've never visited, or visited without a camera.

The author is not a professional photographer or journalist. The text may contain typos, and the photos may have sensor dust spots that I didn't notice. Read more

Today the site turns one week old. Over these days, in the evenings, I've been adding new utilities and sections.

The site now has 5 new utilities and a photo archive section. Here's what was added:

Utilities

In this section, I try to add useful services that help me in my daily work. Here's what's available now:

• Text Comparison — a simple utility that helps find differences between two texts. It shows where and which characters were added or removed.
• Text Case Converter — allows you to convert a string to upper or lower case.
• Epoch Timestamp Converter — converts Unix Timestamp to a readable format and vice versa.
• IP Address Information — a service that shows which country and city an IP address belongs to.
• Deeplink Testing — allows you to test how any deeplinks in your application will work.
• URL Decoder and Encoder — converts a URL to a readable form, or conversely, encodes all special characters.

The list will continue to grow.

Photo Archive

A new section on the site where I will gradually transfer my photo collection.

I've been into photography for fifteen years now. Over this time, I've accumulated many shots — from successful ones to those that don't carry particular artistic value. However, they all allow you to see how our city and other places have changed through the lens of time.

For each shot, I try to add the coordinates of the location where it was taken, the date of the shot, and my thoughts about what's depicted in the photograph. Read more

There are many tools on the internet for comparing two texts. But I wanted to always have one at hand, so the first utility I added to the site was a tool for comparing two texts. The tool turned out to be very simple and clear. And it also works and loads fast. I'd be glad if it comes in handy for you too. https://legost.in/utilities/text-diff Read more

So, by the end of the first day, we have a simple blog that works without a database.

All content is pulled from markdown files. Overall, the implementation turned out to be extremely simple and naive. The architecture chosen for the application itself is a DDD approach with separation into individual modules within a monolith.

I write the posts themselves in the Bear editor. It seemed the most convenient to me, and its free version is more than enough for writing simple posts. Read more

First of all, hello everyone! My name is Vyacheslav, and today marks exactly 10 years since I started earning a living by developing backends for web and mobile applications.

The topic of development interested me since elementary school. In fifth grade, we started learning QBasic and Pascal, and even then I realized that I would dedicate my entire life to this.

I currently work as a team lead at one of the largest IT companies in Kazakhstan, and I enjoy it. But sometimes I want to have the ability to simply express my thoughts, write about what interests me, experiment a little, create some small, possibly useful, possibly fun service and make it publicly available. I want this website to become a platform where I can do all of this.

Why not a blog on a popular platform?

For the same reason — I want to be able to control how everything is displayed and works. Even this page you're reading is a small experiment. The site is built on Laravel, and posts are not stored in a database but rendered from markdown files. This way, I won't have to build an elaborate admin panel for content management, and I won't be limited by built-in editors. The goal is this: the site should be as fast, simple, and hopefully well-indexed as possible.

In addition to blogging, I want to publish simple utilities that I find useful — maybe they'll come in handy for you too. I don't consider myself some great developer or thought leader; I'm simply interested in doing this.

If you're seeing this post, it means that at the very least, the markdown file parsing is working, and the wonderful library that helped me with this is https://github.com/erusev/parsedown Read more