← All sources

TLDR AI news

The latest AI and developer news from TLDR AI, with short summaries.

Visit website →

When LLMs Get Personalized

Even when LLM answers are personalized, they don't become random but retain a shared core.

·2026.04.23 09:00

Microsoft pushes to shift GitHub Copilot subscribers to full token-based billing

Microsoft plans to switch GitHub Copilot to token-based billing starting in June.

·2026.04.23 09:00

Google Unveils Gemini Enterprise Agent Platform to Power Next-Generation Agents

Google has unveiled the Gemini Enterprise Agent Platform, unifying agent building, operation, and governance.

·2026.04.23 09:00

A good AGENTS.md is a model upgrade. A bad AGENTS.md is worse than no docs at all

A well-crafted AGENTS.md boosted code generation quality, but bad documentation actually did harm.

·2026.04.23 09:00

Benchmarking Inference Engines for Agentic Workloads

They released new benchmarks and a harness tailored to the multi-turn, tool-calling patterns of agentic agents.

·2026.04.23 09:00

Advancing Search-Augmented Language Models

Covers directions for improving the performance of language models combined with search.

·2026.04.23 09:00

Former OpenAI Researcher Jerry Tworek Launches Core Automation

Former OpenAI researcher Jerry Tworek has launched Core Automation.

·2026.04.23 09:00

Google Unveils Workspace Intelligence for Gemini Workspace

Google has unveiled Workspace Intelligence for Gemini Workspace.

·2026.04.23 09:00

Qwen3.6-27B: Flagship-Level Coding Performance Achieved with a 27B Dense Model

Qwen3.6-27B is a 27B dense model that claims coding performance surpassing the previous ultra-large MoE flagship.

·2026.04.23 09:00

ChatGPT introduces workspace agents

ChatGPT has unveiled workspace agents for teams to automate repetitive tasks.

·2026.04.23 09:00

Building Agents That Connect to Production Systems Using MCP

Compares API, CLI, and MCP as ways for agents to integrate with external systems, and explains the importance of MCP in cloud environments.

·2026.04.22 09:00

Agent-World Training Arena (3-min read)

Agent-World self-evolves agents using 2,000 environments and 19K+ tools.

·2026.04.22 09:00

Anthropic Developing Always-On Agent with UI Extension Features

Anthropic is developing Conway, an always-on agent with UI extensions and mobile parity.

·2026.04.22 09:00

The Collapse of the Theorem Economy

AI is shaking up theorem production, but the core of mathematics is still understanding.

·2026.04.22 09:00

Deep Research Max: A Leap Forward for Autonomous Research Agents

Google unveiled Deep Research and Deep Research Max, built on Gemini 3.1 Pro to maximize performance and efficiency.

·2026.04.22 09:00

Sam Altman Criticizes Anthropic's Cybersecurity Model Mythos as 'Fear-Based Marketing'

Sam Altman criticized Anthropic's Mythos as fear-based marketing.

·2026.04.22 09:00

OpenAI Pushes to Sell Codex Through Partnerships with Consulting Firms

OpenAI is teaming up with consulting firms like Accenture and Capgemini to win enterprise customers for its AI coding tool, Codex.

·2026.04.22 09:00

Stitch's DESIGN.md format has been open-sourced

A draft specification for Stitch's DESIGN.md has been open-sourced.

·2026.04.22 09:00

CrabTrap: An LLM-as-a-judge HTTP Proxy That Protects Agents in Production

CrabTrap protects production agents by judging their HTTP/HTTPS requests with an LLM.

·2026.04.22 09:00

Neural Networks' Fatal Bits

Flipping just a few sign bits in the weights can effectively collapse vision and language models.

·2026.04.22 09:00