OpenAI Launches Dots Autonomous Agents and GPT-6.1 Sol Model

Launch of Dots Autonomous Agents

OpenAI opened general availability for Dots on September 30, 2026, following its annual DevDay conference in San Francisco. Dots represent an agent architecture designed to run continuously in the background without requiring continuous user prompts.

Each Dot operates inside a dedicated cloud virtual machine equipped with an isolated web browser. This infrastructure allows the agent to conduct proactive research, process documents, and update files while the user is away from the interface.

Dots integrate with over 4,000 external applications through software connectors. The feature is available immediately to ChatGPT Pro and Business Premium subscribers, supporting direct workflows inside ChatGPT, Slack, and Microsoft Teams.

Architecture and Pricing of GPT-6.1 Sol

Alongside Dots, OpenAI released GPT-6.1 Sol, a mid-tier reasoning model designed for agentic coding, computer use, and structured document analysis. The model offers capabilities close to the flagship GPT-6 Astra at roughly one-fifth of the operating cost.

Technical parameters and API pricing for GPT-6.1 Sol include:

• Context window: Supports up to 1,050,000 tokens for text and visual inputs.

• Input pricing: 2.00 dollars per one million tokens.

• Output pricing: 10.00 dollars per one million tokens.

• Cached input pricing: 0.10 dollars per one million cached tokens.

The pricing structure targets production teams that run high-frequency automated loops where flagship tier costs become prohibitive.

Updated Pro Subscription Tiers

OpenAI expanded its Pro subscription options to handle compute allocations for background agents and accelerated reasoning:

• Pro 100: Priced at 100 dollars per month, providing access to one active Dot.

• Pro 200: Priced at 200 dollars per month, offering expanded execution allowances.

• Pro 500: Priced at 500 dollars per month, including the new Ultrafast inference tier.

Dots usage is currently excluded from standard ChatGPT prompt limits. Finalized consumption policies for each tier are scheduled for publication in late October 2026.

Engineering Considerations for Production Workflows

Deploying autonomous agents with persistent cloud browsers introduces specific operational requirements for engineering teams:

• Credential isolation: Integrations connecting agents to internal services should use scoped API tokens restricted to least-privilege permissions.

• Review gates: Code generation and infrastructure modifications initiated by autonomous agents must pass staging pipelines with manual approval gates before deployment.