A total disaster

Summary
In this video, streamer and software engineer ThePrimeagen (on his channel The PrimeTime) reviews and critiques the announcements and controversies surrounding OpenAI DevDay 2026. He breaks down the community backlash against OpenAI cutting the $200/month Pro plan limits in half, examines the new $500/month "Pro 500" plan and Ultrafast tier, benchmarks Ultrafast mode live inside OpenAI Codex, and mocks the on-stage failure of the newly announced "dots" personal agents.

What is shown

• OpenAI DevDay Announcements & Backlash [00:00 - 01:05, 02:06 - 03:35]: Coverage of tweets by OpenAI's Thibault Sottiaux ("Tibo") announcing the halving of effective API/token limits for $200 Pro subscribers, alongside widespread user cancellations on X.
• Sponsor Segment [01:10 - 02:02]: Demonstration of Namespace Devboxes (namespace.so) for isolated agent environments with egress filtering and SSH access.
• Pro 500 Plan & Ultrafast Announcement [03:36 - 04:31]: The slide introducing the $500/month Pro 500 tier offering 25x Plus usage and "Ultrafast" generation speeds (up to 300 tokens/second, 8x faster, 6x more expensive in API).
• Ultrafast Codex Demo & Limit Burn [04:32 - 05:30]: Live terminal test using OpenAI Codex with GPT-6-Astra high ultrafast. Prompting it to implement a pending job in a migration file completes 5 file edits (+722/-16) in 1 minute 41 seconds, burning 2% of the user's total weekly allocation.
• The "dots" Demo Stumble & CFO Gaffe [05:55 - 07:42]: Replay of OpenAI CFO Sarah Friar accidentally calling dots "Muse" on CNBC, followed by the live DevDay keynote demo where presenter Dotty's prompt to dots hangs awkwardly twice before OpenAI tweeted a re-recorded clip blaming "WIFI".
• The dot.com Redirect Troll [07:43 - 08:15]: A clip showing dot.com redirecting directly to SpaceXAI's Grok Bot, reportedly purchased by Elon Musk for $20 million.
• Decisions API, Spaces, Sign-in, and Open Models [08:58 - 12:56]: Overview of other DevDay launches, including the delayed Decisions API, Notion-like "ChatGPT Spaces", "Sign in with ChatGPT" token-sharing, and Codex adding native enterprise support for GLM-5.3 Flash and Kimi K3.

Claims & numbers

• The presenter notes that OpenAI cut the $200/month Pro plan's effective usage allowance in half right before DevDay, claiming models are twice as efficient.
• The new Pro 500 tier costs $500/month, offering 25x the usage of Plus and access to Ultrafast mode.
• Ultrafast token generation claims up to 8x faster output (up to 300 tokens/sec in Codex) while being up to 6x more expensive in the API.
• In the presenter's live test, running one coding task on Ultrafast took 1 minute 41 seconds, changed 5 files (722 insertions, 16 deletions), and consumed 2% of his entire weekly allowance (dropping from 54% to 52%).
• The presenter notes reports that users have burned through their full weekly limit in 40 minutes on Ultrafast.
• Musk/SpaceX reportedly spent $20 million to acquire dot.com to redirect it to Grok Bot.
• Dots usage was initially messaged as unlimited on Astra for Pro users, but clarified to be unlimited only for the first month, with delegated tasks drawing from normal usage limits.

Notable quotes

• [00:33] "Oh by the way, usage? It's going to get cut in half. Happy DevDay!"
• [04:12] "It's named Ultrafast because it's 8 times faster and 6 times more expensive. That's right, that means you can spend your money ultrafast."
• [07:33] "I can't believe they tried to give the old Wi-Fi excuse when it clearly worked on screen like that... Don't play me for a fool."

Assessment
This is an independent commentary and reaction video by a developer reviewing official OpenAI DevDay 2026 footage, company tweets, and keynote demonstrations. The presenter includes a real, unedited screen demonstration of the Ultrafast speed tier running in OpenAI Codex from his own terminal.

Described by gemini-3.8-flash on 2026-10-07 from the video's audio and frames.