Best of OpenAI — February 2026

  1. 1
    Article
    Avatar of bytebytegoByteByteGo·14w

    How OpenAI Scaled to 800 Million Users With Postgres

    OpenAI scaled PostgreSQL to handle millions of queries per second for 800 million ChatGPT users using a single-primary architecture with read replicas. Their approach focused on three pillars: minimizing primary database load through read offloading and write optimization, query and connection optimization using PgBouncer for connection pooling, and preventing cascading failures with cache locking and rate limiting. They addressed PostgreSQL's MVCC constraints by migrating write-heavy workloads to sharded systems and enforcing strict schema change rules. The system achieves five-nines availability with low double-digit millisecond p99 latency through systematic optimization rather than sharding.

  2. 2
    Video
    Avatar of fireshipFireship·14w

    7 AI updates breaking the SaaS business model...

    Recent AI developments are threatening the traditional SaaS business model as major software companies lost $1 trillion in market cap. Seven key AI releases demonstrate this shift: OpenAI's Codex app and 5.3 model, Claude's Opus 4.6, Alibaba's Qwen 3 Coder Next, ZAI's GLM5, Minimax M2.5, GitHub Agent HQ, and Waymo's world model. These tools enable AI agents to replace multiple human seats, offer open-weight alternatives to expensive subscriptions, and automate entire development workflows. The core thesis: when AI intelligence becomes abundant and cheap, the per-seat pricing model that drives SaaS profit margins becomes obsolete.

  3. 3
    Video
    Avatar of t3dotggTheo - t3․gg·16w

    OpenAI just dropped their Cursor killer

    OpenAI released Codeex, a new AI coding tool combining CLI, web app, and desktop application for managing AI agents across projects. Unlike traditional code editors, it provides a GUI for orchestrating multiple parallel coding tasks, with features like work trees, cloud environments, automations, and cross-project thread management. The tool shares history between CLI and app, supports multiple editors, and enables developers to manage several concurrent development tasks simultaneously. While work tree implementation has limitations and environment variable management needs improvement, the orchestration layer represents a shift from direct code editing to agent management workflows.

  4. 4
    Video
    Avatar of ailabs-393AI LABS·16w

    This Is What Clawdbot Was Missing

    OpenClaw (formerly Claudebot/Moltbot) is a self-hosted AI assistant with significant security vulnerabilities and cost concerns. Security issues include credentials stored in plain JSON files, malicious community skills, and prompt injection risks. The architecture sends full conversation context with each query, causing high token costs ($128/month for a single daily cron job) and increasing response times (2-119 seconds). Mitigation strategies include using Docker sandboxing, limiting skill installations, setting API budget alerts, using models with built-in guardrails, and running in isolated environments without sensitive data access.

  5. 5
    Video
    Avatar of aicodekingAICodeKing·16w

    Codex Desktop App + Free GPT-5.2 Codex (Tested): Is OpenAI now copying Conductor,Commander?

    OpenAI launched a desktop app for Codex (GPT-5.2), available free for a month and currently macOS-only. The app provides a graphical interface with features like skills, automations, and work trees, drawing comparisons to Conductor. However, the review highlights numerous UI/UX issues including inconsistent design, buggy interfaces, poor context handling, unintuitive controls (like the plan mode toggle), and problematic VS Code integration that spawns multiple instances. The reviewer suggests competitors like Verdant offer superior agentic interfaces despite OpenAI's resources.

  6. 6
    Article
    Avatar of infoqInfoQ·14w

    OpenAI Scales Single Primary Postgresql to Millions of Queries per Second for ChatGPT

    OpenAI scaled a single-primary PostgreSQL instance to handle millions of queries per second for ChatGPT's 800 million users by deploying nearly 50 geo-distributed read replicas on Azure, optimizing query patterns, and offloading write-heavy workloads to sharded systems like Azure Cosmos DB. Key strategies included connection pooling with PgBouncer, reducing write pressure through application-level tuning, implementing cascading replication to reduce primary load, and isolating critical workloads to maintain low-latency performance under global traffic spikes.

  7. 7
    Article
    Avatar of sdtimesSD Times·14w

    This week in AI updates: GPT-5.3-Codex-Spark, GitHub Agentic Workflows, and more (February 13, 2026)

    OpenAI released GPT-5.3-Codex-Spark, a lightweight coding model delivering 1,000+ tokens per second through a Cerebras partnership. GitHub launched Agentic Workflows for repository automation using plain Markdown descriptions. Google added Automated Reviews to Conductor in Gemini CLI and upgraded Gemini 3 Deep Think mode for improved reasoning. GitHub Copilot testing for .NET reached general availability in Visual Studio 2026. Anthropic raised $30 billion in Series G funding at a $380 billion valuation, with run-rate revenue hitting $14 billion.

  8. 8
    Article
    Avatar of wheresyouredWhere's Your Ed At·15w

    Premium: The Hater's Guide To Microsoft

    Microsoft is spending unsustainably on AI infrastructure while revenue growth stagnates. The company has invested $277 billion in capital expenditures since 2022, primarily on GPUs and data centers, yet Azure's growth has flatlined. Capital expenditures now consume 117% of Azure revenue and 45% of total company revenue, while Intelligent Cloud segment growth has stalled despite massive infrastructure investments. CEO Satya Nadella has overseen 80,000+ layoffs since 2014 while transforming Microsoft from an asset-light software monopolist into an asset-heavy behemoth betting on AI returns that haven't materialized. The company's remaining performance obligations suggest declining future revenue despite paper deals with OpenAI and Anthropic worth hundreds of billions that appear financially impossible to fulfill.

  9. 9
    Article
    Avatar of securityboulevardSecurity Boulevard·12w

    MY TAKE: The Pentagon punished Anthropic for red lines it accepted from OpenAI hours later

    President Trump ordered federal agencies to stop using Anthropic's AI, and Defense Secretary Hegseth labeled the company a national security supply-chain risk. Anthropic's offense was refusing to remove contract clauses prohibiting Claude's use for mass domestic surveillance or fully autonomous weapons. Within hours, OpenAI announced a deal to replace Claude on Pentagon classified networks, with Sam Altman claiming OpenAI holds the same red lines — yet the Pentagon accepted those terms from OpenAI while blacklisting Anthropic for identical ones. The author frames this as the latest in a decade-long pattern: from NSA metadata collection to the Apple-FBI iPhone dispute to now AI model behavioral boundaries, the government's surveillance choke point keeps migrating closer to the layer where judgment and language live. Each cycle compresses faster, leaving less time for public deliberation or governance.

  10. 10
    Video
    Avatar of fknightForrestKnight·15w

    Coding with Opus 4.6 and Codex 5.3 is actually insane

    Claude Opus 4.6 and GPT-5.3 Codex were tested head-to-head across TypeScript, Rust, and Java codebases. Opus 4.6 generally produced better, more reliable code with cleaner first-try compilation, though it's slower. Codex 5.3 is faster and cheaper but sometimes requires iteration to fix errors. Both represent incremental improvements over predecessors, reducing back-and-forth iterations. Opus 4.6 handles larger codebases better with its 1M token context window, while Codex 5.3 excels at speed and cost-effectiveness. Real-world testing showed Opus solving bugs that previous versions couldn't, with more thoughtful implementation choices.

  11. 11
    Article
    Avatar of collectionsCollections·15w

    Official Laravel AI SDK Offers Seamless Integration of AI Capabilities

    Laravel released an official AI SDK in beta that provides a unified API for integrating multiple AI providers (OpenAI, Anthropic, Google Gemini, ElevenLabs) into Laravel applications. The SDK supports text generation, image creation, audio processing, embeddings, vector searches, and streaming responses. It features deep Laravel integration with queues and filesystem operations, automatic provider fallbacks, built-in testing fakes, and support for building AI agents with memory retention and structured outputs. The SDK enables developers to create provider-agnostic, testable AI-powered applications like lead qualification systems, semantic search, and RAG systems.

See all OpenAI archives