DevDay 2026 announcements and developer resources

Models and APIs

  • GPT-6.1 Sol improves coding and computer use, with standard token prices at one-fifth of Astra’s.
  • Ultrafast is a paid speed tier. OpenAI reports up to 8× faster token generation in Codex and 6× in the API. Astra is available now; Sol is coming soon.
  • The Agents API is in public beta, with hosted execution, memory, tools, and multi-agent support.
  • Computer use in the Agents API lets agents operate software through its UI.
  • Decisions API is in limited preview. It uses Luna to classify inputs, route requests, or choose an action from predefined answers.
  • Amazon Bedrock Managed Agents runs OpenAI-powered agents inside AWS.
  • Private Intelligence includes ZDR with Private Safety Processing. A Private Inference preview is planned for this fall.

Codex

  • Codex Cloud runs tasks while your laptop is closed, with access across devices.
  • The CLI update adds two-way voice and an /agents view, plus changes to prompt editing, session resumption, and worktree workflows.
  • Code Review brings summaries, diffs, and questions about changes into the review interface. Automatic reviews can run in the cloud.
  • Codex Security Cloud scans for vulnerabilities, investigates findings, removes duplicates, and prepares fixes. Scans can run on a schedule.
  • Decisions API: Enables real-time decision-making by focusing Luna’s intelligence on a specific set of user-defined questions with finite pre-defined answers. In limited preview.

Plugins and distribution

Dots and collaboration

  • Dots are persistent agents with connected apps and their own cloud computer. Texting is coming soon; teams of dots are a future plan.
  • Specialist dots are an enterprise preview for agents assigned specific responsibilities within an organization.
  • A Microsoft Agent 365 integration is planned for managing specialist dots through Microsoft’s governance and security tools.
  • ChatGPT Space gives teams and agents a shared place for files and project context.
  • Pages are editable documents where teammates and agents can work together, including on charts and interactive tools.
  • Collaborative slides are coming in the next few weeks.
  • Teams and Team Tasks support sharing work and assigning recurring tasks as a team.
  • ChatGPT in Slack and Microsoft Teams responds to @ChatGPT mentions and can use connected tools within those conversations.
  • The Meetings plugin produces notes and action items. It’s in beta on macOS for Pro and Business.
  • Shareable profiles collect your Sites and plugins on one page.

Plans and other updates

  • Pro 500 adds 25× the Plus usage allowance and Ultrafast access.
  • Pro 200 subscriptions are open again. New subscriptions have different allowances from grandfathered plans.
  • OpenAI reported 45% lower API time to first token and over 30% faster tool calls and workflows.
  • An OpenClaw enterprise harness was briefly mentioned, without much detail in the keynote.
  • A worldwide usage reset was announced at the end.

Official recap · Full keynote

can’t find at which subscriptions dots will be available ?

Pro, Business and Enterprise (beta).

Welcome to the community @Boogaav .

You can read more here:

I am asking, coz I am on Pro :sweat_smile:

thanks, checking the docs

@Boogaav this might be the issue:

Dots are rolling out in ChatGPT on web, mobile, and desktop starting today to Pro users in markets excluding the European Economic Area, Switzerland, and the UK

As always EU+UK are last with features (especially since EU AI Act came into play early August).

Not sure if it applies to you, if you are in the EEA or somewhere else?

Lots of issues right now: https://status.openai.com/incidents/01M3Q4RK1SM4EMK445GGPG7C0N

The big win here is cost of cached tokens, which is half of GPT 6 Sol:

RANK MODEL EFFORT TASK RESULT TIME TOKENS API EQ
#1 GPT-6.1-Sol HIGH DIGBENCH P-1 WIN 15m04s 6.3M ~$1.1536
#1 GPT-6.1-Sol HIGH DIGBENCH P-2 WIN 28m32s 9.1M ~$1.6296
#2 GPT-6-Sol HIGH DIGBENCH P-1 WIN 17m43s 4.1M ~$1.3431
#2 GPT-6-Sol HIGH DIGBENCH P-2 WIN 30m13s 10.2M ~$2.8227

Figures from the latest release of Codexometer: track your quota, usage, orchestrate sessions & benchmark models! - you can run these benchmarks for yourself …

Cache retention policy is 30 minutes and you pay +25% for “write”, a service on by default and with controls less exposed in subscription products. Thus, think too long between turns and your input cost with a cache miss jumps from $0.05 to $1.25 again.

Where it can work for you is in multi-turn function-calling agentic use of a growing chat context on the same model signaled by the same cache write key, until opaque compaction breaks it again.

I have to congratulate OpenAI for not reusing a common AI or machine learning word when naming the “Dots” feature.

Can I suggest that instead of ChatGPT “Spaces”, that Enterprise be able to share them in a newly-named “ChatGPT Dot Net application”? :melting_face: