Habr AI
Anthropic and ETH Zurich: a long CLAUDE.md worsens agent performance and raises costs
A study by ETH Zurich across 138 repositories found that long CLAUDE.md and AGENTS.md files often reduce coding-agent success rates while al
Gemini 3.1 Pro outperformed ChatGPT 5.4 and Claude Opus 4.6 in a text generation test
An author-led comparison of three top models across four literary tasks found that Gemini 3.1 Pro handled genre, emotion and brevity better,
Anthropic, OpenAI, and Cursor: Eight Levels of Agent Engineering Maturity
Habr AI broke down eight levels of agent engineering — from tab-complete and context engineering to background agents and autonomous teams t
AMD RX580 ran an LLM locally: how to tame ROCm, Ollama, and get GPU inference
An engineer showed how to make an old AMD RX580 run an LLM reliably with ROCm and Ollama, breaking down false signs of GPU activity, hipMemG
Why Claude 4.6 Isn't Enough Without Context: The Main Blind Spot in LLM Development
Even a strong model like Claude 4.6 loses effectiveness without systematically gathered context: a knowledge base, links between services, a
Developer builds AI-powered news aggregator with MCP, DeepSeek, and Telegram bot
After the drone attack over Dubai, a developer built a multilingual news aggregator: 80+ sources, an MCP server for AI agents, an AI chat, a
BorisovAI tested MoE on an RTX 4090 and showed why perplexity breaks LLM evaluation
BorisovAI tested MoE with plug-in experts on a single RTX 4090 and found that an impressive perplexity score does not guarantee real quality
LLM experiment showed how a model’s “personality” emerges in latent space
An experiment with a modular LLM showed that a separate latent vector can store not only text style, but also stable behavioral traits resem
Nvidia hints at an optical chip that could reshape AI data centers ahead of GTC 2026
Ahead of its March 16, 2026 presentation, Nvidia raised expectations with a promise of “a chip that will shake the world,” and silicon photo
Why DeepMind's AGI advances do not answer the key question of machine consciousness
The author explains why growing computing power and DeepMind's AGI advances do not amount to the emergence of consciousness: intelligence ca
Habr AI on the future of work: how AI and robots could return society to a new antiquity
Habr AI argues that the combination of AI, robotics, and neural implants could do more than reshape the labor market — it could divide socie
Unity showed how to build voice NPCs with memory and world context
A step-by-step guide shows how to build voice NPCs in Unity with a local model, dialog memory, knowledge of the game world, and spoken respo
Harvard: AI is cutting junior hiring, and in three years that could affect the entire industry
Harvard reports a drop in junior hiring after AI adoption, while METR points to growing dependence of experienced developers on AI assistant
DeepSeek and GLM-5 beat Yandex in a test of 34 AI models for managers without VPN
The authors of a large test of management scenarios found that DeepSeek V3.2 and GLM-5, available in Russia without VPN, are markedly strong
Google released Gemini Embedding 2 for multimodal RAG with video, audio, and PDF
Google released Gemini Embedding 2, a model that embeds text, images, video, audio, and PDF into a single space and simplifies building mult
Bitrix24 listed eight common mistakes in developing MCP servers for LLMs
A Bitrix24 developer explained why MCP servers fail on authorization, call chains, poor tool descriptions, testing, security, and context ov
Why Yann LeCun's world model idea does not solve the main crisis in LLM development
After Yann LeCun's departure from Meta, his world model concept is once again being discussed as a path beyond LLMs, but critics argue that
A Physical AI pipeline for SO-101 was assembled on top of ROS2 and LeRobot for 30,000 rubles
An open-source stack based on ROS2 and LeRobot makes it possible to build a full Physical AI pipeline on the low-cost SO-101: teleoperation,
SimpleOne launched SimpleGen — an AI tool for development and deployment on the platform
SimpleOne introduced SimpleGen — an AI tool for generating solutions on its platform: developers only need to prepare a repository, access t
Google AI Ultra: how to turn a subscription into a pool of parallel agents and model consensus
Google AI Ultra is being proposed as the foundation for a multi-agent stack: run parallel Gemini workers, delegate routine tasks, and cross-
Sam Altman and the Pentagon: how military contracts could become insurance for OpenAI
An op-ed on the OpenAI-Pentagon connection argues that military contracts give Sam Altman's company not only money and data, but also protec
OpenAI released GPT-5.4 Pro: new records in ARC-AGI-2, FrontierMath, and logic
OpenAI introduced GPT-5.4 Pro — a model that made sharp gains on difficult benchmarks, solves reasoning tasks better, and handles unconventi
nullClaw on Zig outperformed OpenClaw in memory and startup in local AI agent tests
nullClaw, a lightweight AI runtime on Zig, showed near-instant startup and several times lower memory usage in a local comparison with OpenC
OpenAI's ChatGPT 5.4 beat Claude Opus 4.6 and Gemini 3.1 Pro in a Habr comparison
Habr published a comparison of three flagship models in routine tasks: ChatGPT 5.4 ranked first by total score, Gemini 3.1 Pro was the cheap
Study: Cursor speeds up early development, but later adds to the team's technical debt
A study of Cursor found that the AI assistant sharply accelerates code delivery in the first weeks, but then increases complexity, the numbe
Yandex at AI Dev Day Showed How AI Is Already Changing Development at Avito, Ozon, and T-Bank
At AI Dev Day, companies Yandex, Avito, Ozon, T-Bank, and Sber demonstrated where AI is already accelerating development and where the effec
How AI is changing indie development: it's getting harder for solo developers to compete
The columnist argues that AI has sped up MVP launches, but at the same time raised the barrier to entry, intensified marketing competition,
A company without managers: three traps companies fall into when implementing AI
Of 50 executives at ProIT Fest, only three said AI had actually made decision-making easier — even as companies have fewer and fewer manager
Bitrix24 showed how to add four automation robots to a business portal
Bitrix24 released the next part of its practical series and showed how to integrate four robots into a business portal: for cleaning phone n
Anthropic explained how to build skills for Claude Code and why teams need their own marketplace
Anthropic showed which skills actually work in Claude Code, how to write them without extra noise, and why large teams need their own extens