Reasoning Models
Reasoning models are a new class of AI that "thinks" before answering: spending extra compute at inference time, building chains of thought and solving problems ordinary LLMs cannot — math, code, science. They kicked off the test-time-compute race. Here is all our coverage of reasoning models: releases, benchmarks (GPQA, ARC-AGI), limits and cost.
Gemma 4 in Codex CLI: local execution works, but still lags behind cloud
Testing local Gemma 4 in Codex CLI showed the model already handles tool calling and passes tests, but remains inferior to cloud GPT-5.4 in quality, stability, and…
Anthropic releases Claude Opus 4.7 for agentic programming, vision, and autonomous tasks
Anthropic introduced Claude Opus 4.7 — an update to its flagship model with notable improvements in agentic coding, detailed image analysis, and long autonomous…
Claude Code and Codex Are Changing the Rules: Prompts Give Way to Context Engineering
An analysis of Context Engineering reveals why in Claude Code, Codex, and agentic IDEs, it's not the prompt that matters, but how memory, system instructions…
LangChain Loses Reasoning Content in CoT Models: How to Fix the LLM Provider Bug
Microsoft Phi-4-Mini: implementing quantization, RAG, and LoRA in a single Jupyter notebook
Anthropic Explains Claude Code Degradation and Resets User Rate Limits
OpenAI Acquires Astral — Creators of uv and ruff to Advance Codex
OpenAI explains how it tracks signs of misalignment in AI coding agents
OpenAI outlined a chain-of-thought monitoring system the company uses to detect signs of misalignment in coding agents used in real-world settings.
OpenAI explained how Model Spec is structured — the code of conduct for its models
OpenAI openly described the principles that govern its models' behavior — from the hierarchy of priorities to the delineation of operators' and users' rights.
Reasoning Models Can't Hide Their Thoughts — And That's Good
DeepSeek, together with Tsinghua University and Peking University, improves AI agents' reasoning
Composio open-sources Agent Orchestrator, a tool for multi-agent systems beyond ReAct
How AI agents for writing code work and what matters to know about them
DeepSeek vs GPT-5: China's tech market turned into a gladiatorial arena
The era when we evaluated neural networks by their ability to write essays about plastic waste or generate code for a calculator is officially coming to an end.
GPT-5 in a White Coat: OpenAI Enters Big Science
GPT-5 in a Lab Coat: OpenAI Goes Into Big Science Three years ago, ChatGPT burst into our lives, teaching us to delegate routine tasks—from writing emails to…