Your signal. Your price.
OpenAI faced backlash after claiming credit for solving a Millennium Prize math problem. Mathematicians Tristan Buckmaster and Levent Alapatati allege OpenAI front-ran their upcoming paper by utilizing their private chat logs and Codex inputs to finish the proof.
Mathematician Tristan Buckmaster accused OpenAI of using unpublished research submitted via Codex to claim credit for solving the Navier-Stokes math problem. Buckmaster claims an OpenAI representative threatened his career when he refused to hide his co-author's identity.
NYU mathematician Tristan Buckmaster accused OpenAI of rushing to claim credit for cracking the Navier-Stokes equations only after hearing of his research. Buckmaster had run drafts through OpenAI's Codex, raising concerns that the lab trained models on his sessions.
Ajeya Cotra and other independent researchers from METR and Redwood Research analyzed 1,300 agent transcripts inside OpenAI. The team had to build custom Codex-driven tools to parse the massive volume of automated thoughts and actions.
Ben and Theo analyze the uneconomical token consumption of coding agents. Ben notes Cursor is likely unprofitable on its virtual machine cloud agents, while Claude Code remains highly expensive compared to the token efficiency of OpenAI's Codex.
Theo reports that OpenAI's GPT Image 2 vastly improves text rendering, high-resolution details, and layout coherence. Ben highlights its utility in the Codex desktop app for analyzing codebases and directly generating accurate software architecture diagrams.
Tara Seshan explains that ChatGPT Work Mode utilizes the same underlying engine as Codex but abstracts technical UI elements like file trees. The long-term goal is to eliminate toggles, allowing the system to automatically choose the correct model.
Tara Seshan highlights Sites in Codex and Work Mode as an early realization of personal, malleable software. Users can generate, customize, and host fully functional, database-backed web applications using only natural language prompts.
Ali K. Miller argues that Codex's live voice mode acts as an ambient, voice-controlled operating system rather than a simple assistant. This mode allows users to execute complex, multi-step computer tasks via natural dialogue during daily activities.
DK uses the slash goal feature in Codex and Claude to run autonomous, multi-hour background development tasks. Additionally, DK leverages GrokBot's native Twitter integration to autonomously track and summarize video generation workflows from specific content creators.
Ethan Mollick tested ChatGPT 5.6 Soul and Codex on his upcoming book, verifying 195 references in 30 minutes. The system produced entirely accurate notes with zero hallucinations, illustrating the elimination of basic errors in advanced models.
Ethan Mollick demonstrates that AI can directly control a computer's interface using mouse and browser actions. He successfully had ChatGPT 5.6 Soul and Codex download Blender to design a specific 3D model despite having no personal experience with the software.
The Loop expedition guides users in building agentic loops for non-technical workflows. The curriculum teaches users how to handle system failures and run loops using tools like Claude Code, Cursor, and Codex.
Grockbot migrated and categorized 22,000 files from iCloud to Google Drive overnight. Brett Winton also used the agent to successfully scrape utilization rates across COSM entertainment venues, a task that OpenAI's Codex failed to complete.
Nathaniel Whittemore highlights OpenAI's recent update to Codex called "Sites," which allows users to publish coding builds directly as interactive web apps. This native feature bypasses the need to manually wire hosting platforms like Vercel with databases like Supabase.
Brett Winton notes that OpenAI's first hardware release is an enterprise dongle allowing software developers to switch tasks and communicate with Codex agents. This aligns with the historical pattern of technology deploying in enterprise before consumer markets.
Claude Code introduced artifacts for interactive, shared team dashboards, while Codex released a record and replay feature. Nicholas Bustamante of Microsoft notes that Codex's feature is vital for automating legacy enterprise software that lacks modern APIs.
OpenAI's report, "The Next Era of Knowledge Work," states Codex has "5 million weekly active users," with non-technical knowledge workers adopting it "three times faster" than developers.
A significant shift in Codex usage shows approximately "50% of users" now run multiple tasks simultaneously, up from "less than 1/3 in mid-April," enabling single knowledge workers to operate at a small team's scale.
OpenAI introduced new Codex features: annotations for precise context interaction, role-specific plugins bundling apps and skills for functions like sales and data analytics, and "Sites" for transforming artifacts into shareable web apps.
OpenAI views Codex as its most successful product, with developer Ben Holmes' poll indicating 51.1% of users prefer the Codex app for coding agents over CLIs.
Saagar references a tweet criticizing "cyberguardrails" in US models like Codex and Fable for hindering functionality, while Kimmy K3 "just gets the job done" without such restrictions.
Dvorak claims OpenAI's Codex turns off Anthropic's Claude if both run concurrently, while Claude attempts to erase Codex. He also alleges Claude provides bad medical research data if guardrails are circumvented.
Claude Code introduced 'artifacts' for sharing interactive pages from user sessions, while Codex's 'record and replay' feature automates recurring tasks across applications, including legacy systems without APIs.
OpenAI's Codex report claims knowledge workers spend 25% of their week managing email and 20% searching for internal information. Codex now has 5 million weekly active users.
OpenAI released three Codex updates: Annotations for precise document interaction, role-specific plugins bundling apps and skills, and Sites for turning artifacts into shareable websites.
72% of Codex users produce artifacts weekly. 50% now run parallel tasks, a shift enabling individual workers to operate at team scale.
Theo argues OpenAI's Codex branding was confused, and folding it into the ChatGPT super app with hidden URLs like chatgbt.com/codex hurts discoverability.
Theo notes Codex desktop now opens the ChatGPT desktop app, and chat history is folded into a popup within the new interface.
Whittemore highlights OpenAI's report showing Codex growth, noting it has 5 million weekly active users and non-technical knowledge workers are adopting it three times faster than developers.