→A 60-Year-Open Erdős Problem Was Solved With Help From GPT-5.4 Pro
OpenAI says GPT-5.4 Pro helped solve an Erdős problem open for 60 years. The post names Sebastien Bubeck, Ernest Ryu, and Andrew Mayne, but does not disclose the problem name, proof details, or reproducible conditions.
#Reasoning#OpenAI#Sebastien Bubeck#Ernest Ryu
why featured
Featured · importance 74 · hook + resonance
editor take
OpenAI ties GPT-5.4 Pro to a 60-year Erdős problem, but gives no problem name, proof, or recipe. Math claims need receipts, not podcast framing.
sharp
OpenAI chose the slipperiest phrase here: “with help from GPT-5.4 Pro.” It gives the model credit without saying whether it found the lemma, searched cases, edited prose, or just nudged a human. The disclosed hooks are 60 years, Erdős, Sebastien Bubeck, Ernest Ryu, and Andrew Mayne; the problem name, proof, transcript, and reproducible setup are absent.
Math is the worst place to accept launch-post evidence. DeepMind’s AlphaGeometry at least shipped a task set, method, and contest conditions. This post gives less than an arXiv abstract. GPT-5.4 Pro may have made a real contribution, but the public evidence supports only one claim: OpenAI has a strong story about mathematical research, not yet a verifiable result.
→OpenAI releases upgraded Codex with cross-tool task execution
OpenAI said Codex can now use apps on Mac, connect to more tools, and handle ongoing and repeatable tasks. The post also claims image creation, learning from prior actions, and remembering user preferences; it does not disclose app coverage, integration method, pricing, or rollout timing.
#Agent#Tools#Memory#OpenAI
why featured
Featured · importance 100 · hook + knowledge + resonance
editor take
Codex is no longer pitching autocomplete; it wants the developer’s desktop. The 90+ plugins and macOS computer use are the land grab.
sharp
All four sources orbit the same OpenAI release, with only headline framing diverging: OpenAI says “almost everything,” while Chinese posts sharpen it into “operates your computer.” The hard hooks are concrete: 3 million weekly Codex developers, 90+ plugins, macOS computer use, SSH devbox alpha, gpt-image-1.5, memory, and multi-day automations.
I think OpenAI is making a clean move at the ugly work outside the IDE: PR comments, JIRA, Slack, Gmail, Notion, browsers, terminals. Cursor and Windsurf still fight for the editor surface; Codex is trying to own the software delivery loop. The catch is operational, not demo quality: rollout starts for ChatGPT-signed-in desktop users, while EU/UK and enterprise memory lag. A desktop agent that clicks, types, remembers, and wakes itself up lives or dies on permissions, audit trails, and rollback.
→OpenAI introduces new $100 monthly ChatGPT Pro tier to support growing Codex usage
OpenAI set a new ChatGPT Pro tier at $100/month and raised Codex usage to 5x ChatGPT Plus. The tier keeps all Pro features, including the exclusive Pro model and unlimited Instant and Thinking access. Through May 31, $100 Pro subscribers get up to 10x Plus usage on Codex; the real signal is separate pricing for heavy code-agent demand.
#Code#Tools#OpenAI#Product update
why featured
Featured · importance 88 · hook + knowledge + resonance
editor take
OpenAI adds a $100 Pro tier for Codex growth, but the body gives no quotas; this smells like moving developers off Plus into pricier rent.
sharp
Four sources circle the same OpenAI subscription change, and two are OpenAI posts, so the alignment reads like official seeding: a new $100/month Pro tier, while $200 Pro stays the highest-usage option, with Codex usage as the trigger.
I don’t read this as “more choice.” OpenAI is admitting coding-agent workloads don’t fit cleanly inside Plus economics. The body gives no Codex quota, rate-limit, or Plus downgrade detail, and that gap matters. Cursor and Claude Code have trained developers to run agentic coding as a daily loop, not a novelty. OpenAI’s $100/$200 split is a willingness-to-pay filter before it is a product upgrade.
OpenAI is rolling out ChatGPT in CarPlay to iPhone users on iOS 26.4+ where CarPlay is supported. The post confirms voice mode is available in-car, but does not disclose regions, vehicle coverage, or feature limits. The key shift is distribution into the driving interface, not a new model launch.
#Audio#Tools#OpenAI#ChatGPT
why featured
Featured · importance 74 · hook + resonance
editor take
ChatGPT in CarPlay is OpenAI grabbing the in-car voice slot; only iOS 26.4+ is disclosed, with regions, cars, and permissions left blank.
sharp
OpenAI putting ChatGPT into CarPlay is a distribution move, not a model story. The hard facts are thin: CarPlay support, familiar voice mode, and rollout to iPhone users on iOS 26.4+ where CarPlay works. Regions, vehicle coverage, and access to navigation, messages, calendar, or vehicle controls are not given.
The slot matters because in-car voice has been trapped between Siri, Google Assistant, and automaker systems that still feel command-based. ChatGPT has a better shot at multi-turn conversation, but driving is a brutal environment for latency, hallucination, and accidental actions. Until OpenAI lists the permission surface, I’d treat this as a high-frequency microphone grab rather than a real in-car agent.