Probably the most intelligent right now is Dots, with Astra, but at the cost of speed, and it is not too smart for daily use (reading mail and simple tasks). That is overkill for many subs, but with 6.1 Sol it would sit in the sweet spot. Muse does great work noticing its own potential with video and image generation. Dots will not have video generation, but OpenAI is linking that subscription to all your power tools, which could make Dots great at managing the tools you use and pay for. Grok has voice, video, and image as well, but they are barely integrated: Grok Bot says it cannot generate images and you have to use Grok Build or XAI OAuth. They need to use their own products better. After that, the race is who can bring in more third-party tools. Grok and OpenAI already let you log an AI into third-party tools and harnesses, so it will be tight. Muse makes integrations easy to develop and already has a big audience; a single swap could move the balance. It is close, but Muse and Grok Bot have more potential to lead by owning Instagram, WhatsApp, and X. OpenAI needs to stay relevant and ship big products. Their next hardware could be a game changer.
@elonmusk Can we get grok bot unlimited use for 1 month too for super grok + (100$ ) subs.
Grok bot with 4.7 vs dots with Astra! For common things I believe the speed should be a plus! But if dots would low to 6.1 sol with the speed will be a tough competitor until grok 4.8 or 5!
What’s your opinion about dots and muse?
@OpenAIDevs I already doing this with codex , GitHub and my Hermes in my vps! Even I access to my Hermes vps and my bots with my own iOS app connecting to my Hermes api!
@Ashutosh_Sidh@404notfoundmedi Until your boss put you to develop with Claude you tool for integrate it with OpenAI market, login with OpenAI , and marketing department for run ads on it! 🤣! So you will be out working in OpenAI APIs 🤣
DevDay keepers: Pages — a new document type built for humans + agents.
Write, research, charts, images, visualizations — anything you can do in ChatGPT, on a page. Create it in a conversation, then invite the team. Pairs with Spaces as the collab surface; collaborative slides are coming soon (multiplayer edit + export to PPT / Google Slides).
This is the “productivity that wasn’t designed for AI teammates” gap closing in public.
#ChatGPT #DevDay2026 #Pages
DevDay keepers: ChatGPT showing up where work already happens.
• @ChatGPT in Slack and Microsoft Teams — mention it in a channel/thread/DM; teammates can refine without each needing a license (Business/Enterprise).
• Meetings plugin — notes + personalized summaries/action items into Space; audio deleted when notes are ready (macOS desktop beta for Pro/Business; Enterprise soon).
Less “open another tab,” more “AI in the thread.”
#ChatGPT #DevDay2026 #Productivity
DevDay keepers: Plugins go beyond the chat sidebar.
Sites can now host supported ChatGPT plugins — same app, each teammate’s own connected data/permissions (Business/Enterprise/Edu/Healthcare). And MCP Events let plugins kick off automations when something happens in a connected app (new board task → ChatGPT drafts a plan while you’re away).
The open-ecosystem bet is getting real plumbing.
#MCP #DevDay2026 #ChatGPT
DevDay keepers: Codex tooling refresh (beyond cloud).
Refreshed CLI: start/steer with voice, new /agents view for multi-task delegation, cleaner terminal UI. Code Review in the ChatGPT desktop app — summaries, diffs, ask Codex about issues before PR/MR feedback; automatic first-pass reviews in the cloud while you’re away.
Romain’s UltraFast demos made the “think → ship” loop feel almost unfair. The CLI/review pieces make it daily-driver.
#Codex #DevDay2026 #OpenAI
@sickdotdev If they don’t nerf sonnet it’s ok.. but it’s more probably they nerfe before sonnet to keep it away or update opus to 5.6 sooner that we think!
To be honest if the models are good enough, and they are giving resets.. you will still be getting closer to what used to be.. but keeping the game on resets to make everyone happy, and maybe they are routing better models for task automatically to
Optimize.. Luna is quite good for summarization, translation, etc… but many people won’t set it up or swap models so let’s see… maybe they can do things to still make your tokens more efficient use better the models and still with resets get closer to old tokens but in a way where is sustainable to OpenAI plus a mind game to user to feel more gift!