Quick answer
OpenAI announced more than 20 products at DevDay on September 29, 2026. For developers, the five that matter most are GPT-6.1 Sol ($2/$10 per million tokens, near-Astra quality at one-fifth the price), computer use in the Agents API, the Decisions API, Codex in the cloud, and the Ultrafast speed tier.
Start by testing Sol on the workloads you currently give Astra. That's where the biggest savings are.
DevDay was the biggest OpenAI developer event of the year, and the announcements came fast. The headline consumer product was Dots. The developer story was quieter but bigger: cheaper models, agents that can run software, and a lot of Codex. This recap sticks to what's confirmed. Some details are still rolling out, and I've flagged those. If you're choosing between models right now, see our Opus 5.5 vs GPT-6 Astra guide.
Models and pricing
GPT-6.1 Sol is an upgrade to OpenAI's workhorse model. OpenAI says it gets close to Astra on coding and computer use. The API price is $2 per million input tokens and $10 per million output tokens, the same as the previous Sol and one-fifth of Astra's standard rate (The Decoder, OpenAI Developer Community). The near-Astra claim is OpenAI's own, so test it on your tasks before you move production traffic.
Ultrafast is a paid speed tier. OpenAI reports up to 8x faster generation in Codex (about 300 tokens per second) and up to 6x in the API. API pricing is 6x the standard rate, so GPT-6 Astra Ultrafast runs $60 input and $300 output per million tokens (The Decoder). Astra Ultrafast is available now. Sol Ultrafast is expected in the coming days.
GPT-6.1 Astra is not shipping. Axios reports OpenAI said this week it won't release its newest flagship because of security concerns (Axios).
| Model | Input / output (per 1M tokens) | Status |
|---|---|---|
| GPT-6 Astra | $10 / $50 | Available |
| GPT-6 Astra Ultrafast | $60 / $300 | Available now |
| GPT-6.1 Sol | $2 / $10 | Announced |
| GPT-6.1 Astra | n/a | Not being released |
Sources: The Decoder, OpenAI Developer Community, Axios, published September 29, 2026. Astra base price from the OpenAI model page.
Agents API and Decisions API
The Agents API has been in public beta since September 11. It runs on the same infrastructure as Codex and ChatGPT, with hosted execution, memory, tools, and multi-agent support. The new piece is computer use: agents can now operate software through its UI. OpenAI also says the API now includes Codex's multi-agent features, Tool Search, and Context Compaction, which condenses long contexts automatically (The Decoder).
Amazon Bedrock Managed Agents picks up the same core features, so you can run OpenAI-powered agents entirely inside AWS.
The Decisions API is the sleeper announcement. You define a question with a fixed set of answers, pass in text or images, and get one answer back. Think approve or reject, route to team A or B, or pick an agent's next step. It runs on a version of GPT-6 Luna, and OpenAI says it decides ten times faster than Luna through the regular API. It's a limited preview, with wider availability expected within days. It's a good fit for the cheap, fast half of a two-model setup, where a stronger model plans and a small one picks actions.
Codex updates
- Codex in the cloud: tasks keep running with your laptop closed, and you can start or check them from your phone. Reusable environments let a team share one approved configuration. Available on Plus, Pro, Business, Enterprise, Edu, and Healthcare plans.
- New CLI: two-way voice, an
/agentsview that shows several tasks at once, and better prompt editing, session resuming, and Git worktree handling. Available on all plans. - Code review in the desktop app: summaries, diffs, and questions about changes before you comment on GitHub pull requests or GitLab merge requests. Automatic reviews can do a first pass in the cloud. Available on all plans.
- Codex Security Cloud: scans whole GitHub repos on demand or on a schedule, investigates findings, removes duplicates, and prepares fixes. Available for Pro, Business, Enterprise, and Edu.
Source: The Decoder. If you haven't used it yet, start with our Codex review.
Dots, ChatGPT Space and plugins
Dots are persistent agents with connected apps and their own cloud computer, powered by Astra. Texting them is coming soon. Specialist dots, agents with set responsibilities inside a company, are an enterprise preview, with a planned Microsoft Agent 365 integration for governance.
ChatGPT Space is a shared workspace for teams and agents, with editable Pages, team tasks, and @ChatGPT in Slack and Microsoft Teams. Collaborative slides are due in the next few weeks. A Meetings plugin (beta, macOS, Pro and Business) writes notes and action items.
For builders, plugin extensions can now add sidebar entries, interactive panels, and file viewers, and Figma and Adobe showed integrations. Plugins can also run inside Sites, MCP events can trigger automations, and Sign in with ChatGPT lets people use their ChatGPT subscription usage in 16 launch partners' products (OpenAI Developer Community, Axios).
Plans and privacy
A new Pro 500 plan costs $500 a month, with 25x the Plus usage allowance and Ultrafast access. The existing $200 Pro plan drops to 10x Plus usage in ChatGPT Work and Codex, down from 20x, and GPT-6 Pro chat messages fall from 200 to 100 a week. Current subscribers keep their limits for a transition period and get a one-time credit (The Decoder). Pro 200 is open to new subscribers again, with the new allowances.
Private Intelligence adds zero data retention with Private Safety Processing for businesses. A Private Inference preview is planned for this fall. OpenAI also reported 45% lower API time to first token and over 30% faster tool calls and workflows.
What to do this week
- Bake off Sol against Astra. Pick 10 tasks you currently send to Astra, run them on GPT-6.1 Sol, and compare pass rate and cost. If Sol holds up, you cut price by 80%.
- Prototype one Decisions API step. Replace an LLM routing or classification call once it's generally available.
- Try Codex in the cloud for a long task you'd normally babysit.
- Hold off on Ultrafast unless latency is costing you money. 6x the price is a steep premium.
- Check your plan. If you're on Pro 200, review your new allowances.
Bottom line
The developer takeaway is price and autonomy. Sol brings near-flagship quality down to $2/$10, the Agents API gives you computer use, and Codex now works when you're away. Most benchmark and speed claims are OpenAI's own for now, so verify them on your own workloads. Compare against Anthropic's side in our Opus 5.5 vs GPT-6 Sol, Luna and Grok 4.7 guide.
Frequently Asked Questions
What did OpenAI announce at DevDay 2026?
OpenAI announced more than 20 products on September 29, 2026. The biggest were Dots (always-on agents with their own cloud computer), GPT-6.1 Sol, Codex in the cloud, Codex Security Cloud, computer use in the Agents API, the Decisions API, the Ultrafast speed tier, ChatGPT Space, and a $500 Pro 500 plan.
How much does GPT-6.1 Sol cost?
GPT-6.1 Sol costs $2 per million input tokens and $10 per million output tokens, one-fifth of GPT-6 Astra's standard price and the same as the previous GPT-6 Sol.
What is the Ultrafast tier?
Ultrafast is a paid speed tier. OpenAI says it makes token generation up to 8x faster in Codex and up to 6x faster in the API. API pricing is 6x the standard rate, which works out to $60 input and $300 output per million tokens for GPT-6 Astra. An Ultrafast version of GPT-6.1 Sol is expected soon.
Is the Decisions API available yet?
It launched as a limited preview. The Decoder reports that broader availability is expected in the coming days. It's built on a version of GPT-6 Luna and is meant for classification, routing, and picking from a fixed set of answers.
Is GPT-6.1 Astra available?
No. Axios reports OpenAI said this week it would not release GPT-6.1 Astra because of security concerns. GPT-6 Astra remains available.
Advertiser disclosure: some links on this website are affiliate links, meaning No Code MBA may make a commission if you click through and purchase.