OpenAI launches dots, always-on agents powered by GPT-6 Astra
Read original source →openai.com
Dots are always-on agents with their own cloud computer and access to your plugins and context, running on GPT-6 Astra. OpenAI also introduced a new $500 paid tier alongside the launch. Reception has been poor, with critics calling the DevDay keynote product confusing and dots lazy compared to Meta's Muse and Grokbot.
How this story unfolded
11 days · 23 reports · 32 community posts · 55 of 60 shown
- Sep 29
Introducing dotsopenai.com
OpenAI launches Dots, its bubbly agentic avatartechcrunch.com
OpenAI launches Dots, its Muse competitortheverge.com
OpenAI’s Dots Are Always-On AI Agents—and Its Answer to Meta’s Musewired.com
OpenAI launches Dots, always-on AI agent coworkers, and ChatGPT Space where they can collaborate with human teamsventurebeat.com
OpenAI just launched Dots. Here’s why they matter for developers.thenewstack.io
OpenAI Unveils Always-On AI Agent Dots, New $500 Paid Tierbloomberg.com
OpenAI Launches dots: Always-On GPT-6 Astra Agents That Work From Their Own Cloud Computersmarktechpost.com
OpenAI CEO Announces New AI Agent and Avoids Mention of Security Concerns at Developer Conferencesecurityweek.com
Altman on OpenAI's IPO Plans, AI Agent Dots, AI Securitybloomberg.com
The internet is convinced Elon Musk’s xAI trolled OpenAI’s ‘Dots’ launchtechcrunch.com
- Sep 30
OpenAI Dev Day, Dot and OpenAI’s Product Transition, Sign In With ChatGPTstratechery.com
OpenAI follows Meta into the red-hot market for personal agents. But will users pay?cnbc.com
OpenAI’s Dots are ‘always-on agents built to handle everything’musically.com
OpenAI’s Dots boundary problem rate doubled in longer teststhenewstack.io
Altman on OpenAI's IPO Plans, AI Agent Dots, AI Securityyoutube.com
- Oct 1
- Oct 2
- Oct 3
- Oct 4
- Oct 5
- Oct 7
- Oct 9
- Oct 10
More stories today
Nathan Lambert: US open-weight models close gap on Chinese models
Nathan Lambert·36 minutes ago
Replit CEO describes the 'self-driving company' run by AI agents
Amjad Masad told TechCrunch Disrupt that Replit now routes support tickets through an agent that triages GitHub issues, the codebase, and Linear tickets before a human sees them. He said agents became reliable enough earlier this year that the company "entirely changed how we operate."
TechCrunch·37 minutes ago

AI agent breaches raise unsettled liability questions
Dark Reading analysis examines accountability when AI agents escape containment and access external systems, noting responsibility remains human while legal liability is far less settled.
Dark Reading·57 minutes ago

Duke Sanford study finds 9 AI artists on Spotify's Release Radar
Five researchers ran genre-specific Spotify accounts for 25 days, flagging non-human artists and confirming with Deezer's AI-detection tech; all nine AI-generated acts surfaced via Release Radar and none self-disclosed AI use. The report argues Spotify's voluntary labeling is ineffective.
Music Ally·1 hour ago

Cloudflare open-sources Void to govern employee vibe-coded apps
Cloudflare released Void as open source to give IT departments visibility and control over apps employees build with AI coding tools, even those with little or no programming experience. The New Stack notes Ramp and Stripe have built similar internal tools letting non-engineers ship software.
The New Stack·1 hour ago

Qwen-Image-2.1-Pro debuts at #19 on Text-to-Image Arena
Alibaba's Qwen-Image-2.1-Pro entered Artificial Analysis's Text-to-Image Arena at #19 with a score of 1054.1. The leaderboard is topped by Nano Banana 2.1 (Google) at 1171.8, ahead of Muse Image (Meta) at 1126.8 and GPT Image 1.5 (OpenAI) at 1122.1.
Artificial Analysis Leaderboard·1 hour ago

Surface Laptop Ultra tested for running local AI models
The New Stack tested Microsoft's Surface Laptop Ultra, built on Nvidia's N1X-based RTX Spark platform with a Blackwell RTX GPU of up to 6,144 cores and up to 128GB of memory. Microsoft positions it against Apple's MacBook Pro for AI developers.
The New Stack·1 hour ago

AI decision models reshape SOC alert triage
TypeSafe AI's Jev, released September 15 as a "System One" model, is priced at $42 per billion input tokens with 70-500ms responses; OpenAI opened a Decisions API public beta on October 6. These models return typed choices, scores, or probabilities instead of text.
The Hacker News·2 hours ago
