# AI Catchup > Catch up on AI: tools, practices, and what matters this week A weekly newsletter and content hub covering AI tools, best practices, and the developments that actually matter. AI Catchup publishes sourced AI news, practical guides, comparisons, tutorials, editor's picks, and a weekly newsletter. Article pages include canonical URLs, direct-answer summaries, publish and update dates, author information, primary-source links, and structured data. ## Explore - [Tools & Software](https://aicatchup.com/tools): Reviews, roundups, and guides for AI coding tools, writing tools, image tools, and MCP servers - [Practices & Workflows](https://aicatchup.com/practices): How people use AI day-to-day. Prompt engineering, skill files, automation patterns - [News & Analysis](https://aicatchup.com/news): Curated takes on AI industry news with perspective - [Skills & Learning](https://aicatchup.com/skills): Tutorials and guides for leveling up with AI tools - [Comparisons & Roundups](https://aicatchup.com/comparisons): Head-to-head tool comparisons and monthly best-of posts - [Editor's Picks](https://aicatchup.com/picks): Current ranked recommendations - [Newsletter Archive](https://aicatchup.com/newsletter): Weekly AI Catchup editions - [Search](https://aicatchup.com/search): Search the complete content archive - [Topic Index](https://aicatchup.com/topics): Browse articles by product, model, workflow, and technology ## Featured Evergreen Guides - [Ramp Glass Playbook: Internal AI Workspaces for Company-Wide AI Adoption](https://aicatchup.com/practices/internal-ai-workspaces-playbook): Glass is Ramp's internal AI productivity suite, built after the company had already hit 99 percent AI adoption and found most employees still stuck on setup. One Okta SSO sign-in wires up every tool, a marketplace called Dojo holds 350+ shared skills, memory is mined from Slack, Notion, Linear and Calendar, and jobs run on cron. Here is the playbook other companies can borrow. - [Prompt Engineering in 2026: The Playbook That Works Across Claude and GPT](https://aicatchup.com/practices/prompt-engineering-guide): Prompt engineering in 2026 has two layers: cross-model fundamentals (clear direct instructions, motivated rules, 3-5 structured examples, XML boundaries, long documents before the question) and model-specific overrides that matter more than ever, because the newest models need instructions deleted as often as added. Start here, then apply the guide for your model. - [How to Build Your Own MCP Server: A Beginner's Guide](https://aicatchup.com/skills/building-mcp-servers): Building an MCP server involves three steps: scaffold a TypeScript project with the MCP SDK, define tools that expose your desired functionality, and test the server locally with Claude Code or another MCP client. The entire process takes under an hour for a basic server. - [Codex CLI vs Claude Code vs Cursor: 2026 Architecture Deep-Dive (Sandboxing, Context, Plugins, Scheduling)](https://aicatchup.com/comparisons/codex-cli-vs-claude-code-vs-cursor-architecture): Codex CLI, Claude Code, and Cursor all reach for the same outcome -- an AI agent that ships code -- with three different architectures. Codex enforces safety at the OS kernel layer. Claude Code uses application-layer hooks for programmable governance. Cursor builds the agent into a visual IDE with a marketplace of plugins. The right one depends less on which model you prefer and more on which architecture matches your security posture, your composition needs, and your team shape. - [Perplexity Personal Computer vs OpenAI Codex Computer Use vs Claude Computer Use: Which AI Should Run Your Mac?](https://aicatchup.com/comparisons/perplexity-vs-codex-vs-claude-computer-use): Perplexity Personal Computer is the right pick if you want an always-on Mac mini agent and you already pay for Perplexity Max. OpenAI Codex Computer Use is the right pick if you live in ChatGPT and need one agent driving your code, your browser, and your desktop apps on either macOS or Windows. Claude Computer Use is the right pick if you want raw computer use through the API on any OS, or the built-in version in Claude Code and Claude Desktop on macOS or Windows with a Pro or Max plan, as of September 2026. - [Gemini in 2026: Google's Full AI Stack Explained (Chrome, Workspace, CLI, AI Mode, Antigravity)](https://aicatchup.com/tools/gemini-google-ai-stack-2026): Google's AI stack in April 2026 is six distinct products built on the same Gemini model family: Gemini 3.1 Pro for the model itself, Gemini in Chrome and Workspace for consumer surfaces, Gemini CLI for the terminal, Antigravity for agentic IDE work, AI Mode and AI Overviews for Search, and the Gemini API for developers. This guide maps the whole stack, shows where each piece actually wins versus the competition, and tells you which one to reach for in which workflow. ## Latest Articles - [Claude Science Beta: An AI Workbench for Reproducible Research](https://aicatchup.com/news/claude-science-beta): Claude Science is Anthropic's beta research app, not a model. It runs Python and R analyses in a local sandbox, queries 60+ scientific databases, submits jobs over SSH or to Modal, and attaches code, environment, and conversation provenance to every artifact. As of September 2026 it runs on macOS, Windows, and Linux for Pro, Max, Team, and Enterprise plans. - [Ox Alpha Was Z.ai's GLM-5.3-Flash, and the Free Week Is Over](https://aicatchup.com/news/ox-alpha-stealth-model-free-openrouter-opencode): Ox Alpha was Z.ai's GLM-5.3-Flash. OpenRouter's model page now names the developer, the stealth listing serves no providers, and OpenCode has dropped Ox Alpha Free. The named model lists at $0.15 per 1M input tokens and $0.50 output; the 50% launch discount ended September 9, 2026, and as of September 10 Z.ai's pricing page carries list prices only. - [Claude Code +50% Weekly Limits Promo Ended September 13, 2026; Limits Are Now 25% Higher Permanently](https://aicatchup.com/news/claude-code-weekly-limits-50-percent-promo): Anthropic's +50% Claude Code weekly-limits promotion ended September 13, 2026 at 11:59 PM PT. On September 14 Anthropic rewrote its support article: starting September 14, 2026, weekly limits in Claude Code are 25% higher than they were before the promotion for Pro, Max, Team, and seat-based Enterprise plans. Five-hour limits are unchanged. - [Claude Code Weekly Limits Are Now 25% Higher Permanently: Anthropic Confirms the September 14 Change](https://aicatchup.com/news/claude-code-weekly-limits-permanent-25-percent-september-2026): As of September 14, 2026, Anthropic documents the permanent increase on its own support article: weekly limits in Claude Code are 25% higher than they were before the May-September promotion for Pro, Max, Team, and seat-based Enterprise plans. The 50% promotion ended September 13 at 11:59 PM PT. Five-hour limits, prices, and other Claude products are unchanged. - [Claude Managed Agents Adds a Terminal Session Viewer and Auto Permission Mode](https://aicatchup.com/news/claude-managed-agents-session-viewer-auto-mode): Claude Managed Agents now includes a terminal workflow for attaching to live sessions and a local web session viewer. Anthropic also added an auto permission mode that evaluates each agent or MCP tool call against the intent in user.message and runs, denies, or pauses for approval. - [Master Claude Code's 1M Context Window: Rewind, Compact, Clear, and Subagents](https://aicatchup.com/practices/claude-code-context-management): Claude Code's 1M token context window opens longer autonomous sessions but introduces 'context rot' -- degraded performance as the window fills. Master four turn-end tools: /rewind to drop bad branches, /compact to summarize and continue, /clear to start fresh with a distilled brief, and subagents to wall off noisy work in their own context. - [OpenAI Launches ChatGPT for Financial Services](https://aicatchup.com/news/openai-chatgpt-financial-services): OpenAI introduced ChatGPT for Financial Services, a tailored ChatGPT Work experience for eligible financial institutions. It combines hosted datasets from providers such as Daloopa, PitchBook, LSEG News, and Crunchbase with GPT-6 Astra, granular citations, spreadsheet and presentation generation, firm templates, and enterprise access controls. - [ChatGPT Sites Adds Teammate Editing, External Viewers, and Public Publishing Controls](https://aicatchup.com/news/chatgpt-sites-teammate-editors): OpenAI’s latest ChatGPT Sites update expands the sharing model around teammate editing, named external viewers, and public publishing. A Site owner can give an active member of the same ChatGPT workspace Can edit access; editors can update and save the Site, then publish later versions to the same URL after the owner’s first publish. Owners retain access, URL, secrets, custom-domain, and audience controls. - [OpenAI Welcomes the Git AI Team Behind an Open-Source Agent Attribution Tool](https://aicatchup.com/news/openai-welcomes-git-ai-agent-attribution): OpenAI says Aidan Cunniffe and Sasha Varlamov from the Git AI team have joined OpenAI. Their open-source Git extension tracks AI-generated code at line level, linking it to the agent, model, and prompts that produced it, with commands for attribution stats and AI-aware blame. - [Prompting Claude Opus 5: Trim the Verbosity, Delete the Verification](https://aicatchup.com/practices/prompting-claude-opus-5): Claude Opus 5 needs opposite prompting from its predecessors: you prompt for conciseness because effort no longer controls visible length, and you delete verification and double-check instructions because the model already does both. Constrain scope on narrow tasks, cap subagent spawning for cost, and keep thinking enabled at low effort rather than disabling it. - [Claude Code Adds Plugin Evals for Regression Testing and No-Plugin Baselines](https://aicatchup.com/news/claude-code-plugin-evals-regression-testing): Claude Code now includes `claude plugin eval`, a testing workflow for plugin and skill authors. The command can create cases and graders, run a plugin in isolated non-interactive sessions, compare results with a no-plugin baseline, generate an HTML report, and support CI gates for plugin changes. Evals require Claude Code v2.1.269 or later and consume model usage. - [GPT-Rosalind Leaves Research Preview for Eligible Organizations Worldwide](https://aicatchup.com/news/openai-gpt-rosalind-global-trusted-access): GPT-Rosalind, OpenAI's purpose-built life sciences reasoning model, is out of research preview for eligible organizations worldwide through OpenAI's trusted-access program. OpenAI says qualified customers can use it in the API, Codex, and ChatGPT Enterprise, while a Life Sciences research plugin connects Codex to more than 50 public tools and data sources. Published pricing takes effect October 5, 2026. - [Claude Platform's 'ant' CLI Brings the Full Claude API to Your Terminal](https://aicatchup.com/news/claude-platform-ant-cli-managed-agents-worker): Anthropic's 'ant' CLI exposes every Claude API resource as a terminal subcommand. As of September 2026 you install it through Homebrew, curl, or Go, authenticate by browser login, API key, or Workload Identity Federation, manage agents as code with 'ant apply' (CLI 1.30.0 or later), and run a self-hosted Managed Agents worker with 'ant beta:worker poll' or 'ant beta:worker run'. - [Cursor Projects Give a Coordinator Agent a Long-Lived Body of Work](https://aicatchup.com/news/cursor-projects-coordinator-agent): Cursor's Projects beta moves beyond one chat at a time: a coordinator agent keeps context over months, delegates implementation and testing to subagents, shares research and artifacts across cloud and local machines, and can watch PRs, Slack, or a schedule for recurring work. - [OpenAI Opens Its Codex-Powered Agents API to All Developers in Public Beta](https://aicatchup.com/news/openai-agents-api-cloud-agents-public-beta): OpenAI's Agents API is now in public beta for all developers. The managed service uses the Codex harness to run long-lived cloud agents, coordinate parallel subagents, compact context across long sessions, and work in OpenAI-hosted, VPC, or partner-provided sandboxes. - [Claude Code Desktop Adds Pop-Out Panes for Diff and Terminal Work](https://aicatchup.com/news/claude-code-desktop-pop-out-panes): ClaudeDevs says Claude Code Desktop can now pop out any pane into its own window. You can move a diff or terminal to another screen while Claude keeps working, dock it back later, and run sessions side by side or stacked. - [Claude Fable 5.1 and Mythos 5.1: Same Price, Cheaper Cache Reads, Fewer Safeguard Interventions](https://aicatchup.com/news/claude-fable-5-1-mythos-5-1): Anthropic released Claude Fable 5.1, available today as claude-fable-5-1 at $10 per million input and $50 per million output tokens, with cache reads cut 75% to $0.25 per million. Anthropic estimates typical workloads cost about 25% less than Fable 5. Claude Mythos 5.1 is the same model with fewer cyber and life-science safeguards, limited to vetted US organizations. - [OpenAI Shares a Defense Factory Playbook for Agentic Cyber Defense](https://aicatchup.com/news/openai-defense-factory-playbook): OpenAI is sharing a Defense Factory playbook that connects security tools, skills, reproducible environments, and agents into a continuous loop for discovering, validating, assigning, remediating, and independently verifying vulnerabilities. - [Claude Code Desktop Adds Auto-Continue After Usage Limits](https://aicatchup.com/news/claude-code-desktop-auto-continue): Claude Code waits in the open session and continues your task automatically when a claude.ai usage limit resets. It is on by default from version 2.1.234 in interactive sessions signed in with a claude.ai subscription. Turn it off in /config under 'Continue automatically at usage limit' or set autoContinueAtUsageLimit to false. It resumes work; it does not raise the limit. - [Claude Marketplace Adds Cursor, CrowdStrike, Factory, Gamma, and Vercel](https://aicatchup.com/news/claude-marketplace-new-enterprise-partners): Anthropic is expanding Claude Marketplace with Cursor, CrowdStrike, Factory, Gamma, and Vercel. Enterprise customers can use an existing Anthropic spend commitment to buy Claude-powered solutions from marketplace partners, with Anthropic directing buyers to their account team rather than a self-serve checkout. ## Popular Topics - [2026 Trends](https://aicatchup.com/topic/2026-trends): 69 articles - [OpenAI](https://aicatchup.com/topic/openai): 60 articles - [Developer Tools](https://aicatchup.com/topic/developer-tools): 55 articles - [Anthropic](https://aicatchup.com/topic/anthropic): 54 articles - [AI Coding Tools](https://aicatchup.com/topic/ai-coding-tools): 48 articles - [Claude](https://aicatchup.com/topic/claude): 39 articles - [Codex](https://aicatchup.com/topic/codex): 39 articles - [Claude Code](https://aicatchup.com/topic/claude-code): 38 articles - [Cursor](https://aicatchup.com/topic/cursor): 36 articles - [ChatGPT](https://aicatchup.com/topic/chatgpt): 27 articles - [Agents](https://aicatchup.com/topic/agents): 22 articles - [Workflow](https://aicatchup.com/topic/workflow): 17 articles - [AI Agents](https://aicatchup.com/topic/ai-agents): 15 articles - [Computer Use](https://aicatchup.com/topic/computer-use): 12 articles - [MCP](https://aicatchup.com/topic/mcp): 12 articles - [GPT-5.6](https://aicatchup.com/topic/gpt-5-6): 11 articles - [Security](https://aicatchup.com/topic/security): 11 articles - [Enterprise](https://aicatchup.com/topic/enterprise): 10 articles - [Pricing](https://aicatchup.com/topic/pricing): 10 articles - [Agentic Coding](https://aicatchup.com/topic/agentic-coding): 9 articles - [API](https://aicatchup.com/topic/api): 8 articles - [Plugins](https://aicatchup.com/topic/plugins): 8 articles - [Cloud Agents](https://aicatchup.com/topic/cloud-agents): 7 articles - [Automation](https://aicatchup.com/topic/automation): 6 articles ## Machine-Readable Resources - [Complete content catalog](https://aicatchup.com/llms-full.txt): Every article, newsletter edition, and picks page with metadata - [RSS feed](https://aicatchup.com/rss.xml): Latest articles and newsletter editions - [XML sitemap](https://aicatchup.com/sitemap.xml): Canonical indexable URLs with freshness dates ## Publisher - [About AI Catchup](https://aicatchup.com/about) - [Editorial Policy](https://aicatchup.com/editorial-policy) - [Contact](https://aicatchup.com/contact)