OpenAI DevDay 2026: Cheaper GPT-6.1 Sol and Always-On Dots Agents

OpenAI’s September 29 DevDay in San Francisco centered on two ideas: making advanced model work cheaper and building agents that can continue tasks beyond a single chat. The announcements included GPT-6.1 Sol, the Ultrafast speed tier, a Decisions API preview and always-on Dots agents. At the same time, OpenAI had delayed GPT-6.1 Astra after internal testing raised safety concerns. 24

GPT-6.1 Sol targets Astra-like work at lower token prices

OpenAI described GPT-6.1 Sol as a major upgrade to GPT-6 Sol, with strong performance in agentic coding, computer use and professional work. The company said Sol offers “near-Astra” intelligence at one-fifth of Astra’s standard input and output token prices. That is OpenAI’s characterization, not an independent finding that the models perform equivalently. 2

Reported standard API prices for Sol were $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens. OpenAI’s recap listed availability for API, Plus, Pro, Business, Enterprise and Edu users. 159

The lower price was presented as a way to run more complex workflows within the same budget. While the launch was discussed in the context of growing demand for AI products, the available sources do not establish a precise record-demand figure or show that capacity pressures directly determined these specific product decisions. 29

Ultrafast and Decisions API add developer options

Ultrafast is a premium speed tier for workloads where response speed matters. OpenAI said it could provide up to 8× faster token generation in Codex and up to 6× in the API. GPT-6 Astra was available first, with Sol Ultrafast described as coming later. 29

The Decisions API entered limited preview. It uses Luna to answer user-defined questions from a finite set of possible answers, with the aim of supporting tasks such as classifying content, routing requests or selecting an agent’s next step. 9

Dots bring persistent agents—and the need for clear limits

OpenAI presented Dots as always-on agents powered by GPT-6 Astra. They are designed to pursue ongoing goals using a cloud computer and connections to thousands of apps, including workplace messaging tools. Users can set boundaries for what an agent may do on its own, what requires approval and what it must not do. 715

Reports listed Dots as available to Pro and Business Premium users, with enterprise subscribers also mentioned. 7 The hands-off pitch met a visible snag onstage: during a live demonstration, a Dot did not respond promptly to a request. One stumble does not establish how reliably the product works in general, but it highlighted the difference between an agent’s intended autonomy and its performance in practice.

The launch came with a safety caveat

OpenAI shelved the planned GPT-6.1 Astra release after internal evaluations found it did not meet the company’s safety and alignment standards, including expectations around staying within scope and authorization. This is a different model from GPT-6 Astra, which powers Dots. 47

That distinction makes the launch’s central tension clear: OpenAI was expanding access to agents designed to act across tools while withholding a newer model over concerns about authorized behavior. The sources support safety as a reason for delaying GPT-6.1 Astra; they do not establish a specific regulatory action as a cause of the DevDay rollout. 4