#LLM
Curated coverage on «LLM»: releases, research, benchmarks, open-source.
Small Language Models in Agents: Five Use Cases Reshaping AI System Architecture
Five concrete scenarios where SLMs are already replacing frontier models in agents: from task routing and local inference to drafting for GPT-4—with metrics and…
MCP and AI Agent Security: How the Protocol Created a New Attack Surface
Model Context Protocol standardized AI agent connections to tools — but each MCP server is now a trust point, vulnerable to tool poisoning and prompt injection.
How to build an AI harness for SaaS development: combining Orca, Pi Agent, and GitHub
A developer on Habr explains why a well-built harness around a model matters more than the choice between GPT or Claude itself — and shows his working setup.
Meta Creates Cloud Business to Sell AI Capacity, Fable 5 Enters Global Market
Former Qwen Technical Lead: Why Hybrid Reasoning Failed and What's Next
Modal releases a GPU calculator: serverless wins when peak demand exceeds the reservation discount
Anthropic could raise $50B: investors value company at $900 billion
DeepSeek introduces double tariff during peak hours — first in China's AI price war
DeepSeek announced double pricing during peak load hours for V4 model API clients — an unexpected move from the company that itself sparked the AI price war.
Proton updates private AI chatbot Lumo to version 2.0 with new capabilities
On June 30, 2026, Proton released Lumo 2.0 — an updated private AI chatbot with expanded capabilities for users who prioritize data confidentiality.
SimpleOne explained the architecture of AI agents in its enterprise GenAI platform
Sberbank Converted GigaChat3 to Diffusion Mode: How the GFusion Project Works
Company X launched an MCP server for directly connecting AI applications to the API
Anthropic launched Claude Science — a unified environment for scientists instead of a scattered set of tools
Prompt injection via Claude Code skills: how to check a plugin before installing it
OSS skills for AI assistants can contain executable code with user privileges. We explain why prompt injection via plugins is more dangerous than ordinary…
Cursor moves to SpaceX: will the editor retain support for OpenAI and Anthropic models?
SpaceX is buying Cursor — an AI code editor with an audience of millions of developers — and wants to preserve support for third-party models from OpenAI and…