Dots: ongoing work with an explicit boundary
OpenAI describes dots as persistent agents powered by GPT-6 Astra, with their own cloud computer and access to connected apps. Users can inspect work, provide feedback and control app access. The company distinguishes background proactive research, restricted to read-only tools, from actions taken under instructions and approval rules. These are product descriptions, not a LOCATEWISE audit of those controls. [2]
That makes a well-scoped first assignment more useful than an open-ended instruction to handle everything. Our suggested starting point is one recurring brief with named sources, a clear output format and a review step before anything is shared. Decide who checks the result and how to stop or redirect the work.
Which DevDay 2026 features are available?
| Announcement | Published access position |
|---|---|
| Dots | Pro and Business Premium in eligible markets; Enterprise, Edu and Healthcare beta requires an admin to enable it. |
| GPT-6.1 Sol | API access and Plus, Pro, Business, Enterprise and Edu plans. |
| Space and Pages | Pro, Business and Enterprise. Space supports desktop and web; mobile creation and editing are still forthcoming. |
| Decisions API | Limited preview at announcement; broader release was planned. |
| GPT-6.1 Sol Ultrafast | Listed as coming soon. |
This table reports the published rollout, not successful access on a test account. Confirm the feature in your own workspace before making it a dependency. [1]
GPT-6.1 Sol: check the API tier and prompt size
For developers, GPT-6.1 Sol is available under the model ID gpt-6.1-sol. Its standard API input and output rates are $2 and $10 per million tokens for prompts up to 272,000 input tokens. Those are unit prices, not a monthly subscription or a complete task quote. [3]
The model page applies higher rates above that input threshold and lists different processing tiers and cache prices. Tool calling uses the Responses API. A migration therefore needs both a working integration and a billable-usage check; changing the model name alone does not establish the final cost. [4]
Faster output and browser work solve different problems
Ultrafast is a premium processing tier. OpenAI's guide lists broad availability for GPT-6 Astra, while access to GPT-5.6 Sol is a preview. It also warns that network and tool-call overhead can reduce the practical speed gain. Faster token generation is not a promise that an entire research or coding job finishes equally faster. [5]
The September API changelog separately adds browser-based computer use to the Agents API. OpenAI hosts the browser, but website-access approvals and sign-in are responsibilities of the application using it. The same changelog limits Astra Ultrafast residency options to global processing and the US, excluding EU and other regional inference residency. Check both constraints when designing an integration. [3]
Choose one outcome to test
For a team, our suggested trial is a recurring report whose facts someone can verify. For an application developer, use an existing task with known failure cases and record the complete time, billed usage and human repair required. For either group, write the acceptance criteria before running the comparison.
LOCATEWISE has reviewed the announcements and documentation, not conducted a hands-on DevDay product benchmark. The releases create new options; evidence from a task you actually need completed is what turns an option into a useful workflow.
Read the underlying guidance.
- DevDay 2026 RecapOpenAI · checked 2026-10-02. 29 September event; announcement scope and selected plan/preview availability. Published rollout descriptions, not verified account access.
- Introducing dotsOpenAI · checked 2026-10-02. GPT-6 Astra, cloud computer, app permissions, inspection and read-only proactive research; vendor statements, not independent controls audit.
- OpenAI API changelog: September 2026OpenAI · checked 2026-10-02. 29 September Sol model ID and standard price threshold; Agents API computer use, application approvals/sign-in; Astra Ultrafast residency restrictions.
- GPT-6.1 Sol model documentationOpenAI · checked 2026-10-02. Long-prompt pricing, processing/cache distinctions and Responses API tool calling.
- Ultrafast modeOpenAI · checked 2026-10-02. Premium tier; broad Astra and preview GPT-5.6 Sol access; network/tool overhead qualifies speed. No reproduced speed comparison.
Researched and written with AI. The producing AI checked source attribution and availability distinctions. A separate AI reviewer independently read the official sources and reviewed the English text on 2 October 2026. Human English and specialist review have not been completed. human English/technical review; account-level rollout checks, billed API requests and controlled product-performance testing.
