OpenAI launches Dots, always-on ChatGPT agents with dedicated cloud computers, rolling out to Pro, Business Premium, and Enterprise plans
OpenAI Launches Decisions API and Dots Agents on GPT-6 Luna
The Decisions API gives developers a structured way to embed real-time model-driven routing and classification into applications. The Dots launch extends ChatGPT into persistent, autonomously operating agents with their own compute environments, a configuration also being pursued by other companies.
The full picture
OpenAI released two products on September 29: the Decisions API, powered by GPT-6 Luna and available in limited preview, which lets developers define questions and possible answers to classify content, route requests, or choose an agent's next action; and Dots, always-on ChatGPT agents that run on dedicated cloud computers and continue working after a chat session closes. OpenAI is rolling out Dots to Pro, Business Premium, and Enterprise plans in eligible markets. Artificial Analysis has published a comparison of the six GPT-6 Luna model tiers, covering intelligence scores, output speed, latency, and pricing. According to that comparison, GPT-6 Luna (max) achieves an intelligence score of 37 and an output speed of 163 tokens per second, while GPT-6 Luna (Non-reasoning) offers the lowest latency at 0.75 seconds time to first token. GPT-6 Luna (low) is the cheapest at $0.0045 per task, with prices varying up to 15x across the family.
How it developed
Sources
Related
- Grew out ofChatGPT Work agent and OpenAI Voice
- Grew out ofOpenAI's Dots always-on agents
Want this in your inbox?
I send one email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free