OpenAI DevDay Blows the Roof Off: dots Is Born, GPT-6.1 Sol Crashes 80% in Price, and a $500 Tier Arrives
20 announcements in one hour: dots, a 7×24 digital employee with its own cloud computer and 4,000+ tools; GPT-6.1 Sol matching Astra at 1/5 the price ($2/$10, cache $0.10); Ultrafast at 300 tokens/s; ChatGPT Space; Codex fully in the cloud; and the return of $200 Pro plus a $500/month Pro 500 tier.

OpenAI DevDay Blows the Roof Off: 7×24 Digital Employee "dots" Is Born, GPT-6.1 Sol Crashes 80% in Price, and a $500 Tier Arrives
| Channel | URL | Best for |
|---|---|---|
| High-speed GPT mirror in China | https://trygpt.asia/list/#/home | Browser-based chat without a VPN, lightweight workflow testing |
| MomoAI monthly plans | https://momoai.asia/home | Heavy users who want a monthly subscription |
| All-model API aggregator | https://tryallapi.com/ | Developer calls, low-latency direct connection, one-stop multi-model access |
| Official top-up channel | https://chongzhi.trygpt.asia/ | Removing the overseas payment barrier, official direct top-up |
At Fort Mason in San Francisco, OpenAI DevDay absolutely detonated.
Just now, Sam Altman walked onto the DevDay stage and, in a single hour, dropped 20 major announcements back to back!

A brand-new AI species called dots — an AI that stays online 24/7 — is officially born!
Unlike traditional ChatGPT, dots is OpenAI's first "resident digital employee" with its own independent cloud computer, able to run code and watch over projects for you anytime, anywhere.

Even more striking: Altman rolled out the next-gen flagship GPT-6.1 Sol, with Astra-class reasoning and coding ability — at a price that drops 80%.
At the same time, the Ultrafast inference mode went live, boasting a terrifying 300 tokens/s generation speed — 8x the standard tier!

On stage, Altman not only revived the $200/month Pro subscription, he also threw out the jaw-dropping "noble tier" — Pro 500 ($500/month).
Its usage allowance is 25x that of ChatGPT Plus, includes native access to Ultrafast, and eliminates the 5-hour limit entirely!

On top of that, the keynote also shipped ChatGPT Space, the Agents API, Codex Cloud and much more — an almost suffocating amount of information.
At the very end, "Godfather" Tibo wheeled out a giant reset button and led the whole room in a "one-click restart" celebration!

One netizen quipped: "The best and most down-to-earth moment of the entire keynote was that one-click restart segment!"

🔥 DevDay Key Announcements at a Glance
| Announcement | Highlight / breakthrough metric | Who it's for |
|---|---|---|
| dots 7×24 agent | Independent cloud computer, native access to 4,000+ tools, resident in Slack/Teams, no babysitting | Pro / Business Premium / Enterprise |
| GPT-6.1 Sol | Matches Astra, beats Opus 5.5, at just 1/5 of Astra's price | All platforms / API developers |
| Ultrafast mode | 300 tokens/s, an 800% speed-up | Astra / Sol / Pro 500 users |
| Pro 500 | $500/month, 25x Plus allowance, unlimited Ultrafast | Hardcore power users / managers & devs |
| ChatGPT Space | AI-native collaborative workspace with shared team context and real-time interaction | Team collaboration / Enterprise |
| Codex Cloud | Fully cloud deployment, voice control, /agents view, repo-wide automated vulnerability review |
Engineers / dev teams |
🤖 dots Arrives: OpenAI's First Always-On 7×24 AI
The undisputed star of this keynote is dots.
It is absolutely not another chat box, and not some side plugin bolted onto ChatGPT.

As previously leaked, dots is a super-agent that stays online 24/7, powered under the hood by GPT-6 Astra.
It has its own cloud computer and browser, natively wired into 4,000+ mainstream software tools, and can take up residence inside workplace apps like Slack and Teams.
You give it a goal; it decomposes and executes autonomously — no step-by-step hand-holding required. Even with your screen off, dots keeps running tasks in the background around the clock.
You don't need to log into ChatGPT — dots is working for you wherever you are.
The first-run experience is full of ceremony too:
Open ChatGPT and your dot greets you: "Hi, I'm your dot. Just checked in at ChatGPT — from now on, leave whatever comes up to me."

In the live demo, an OpenAI researcher named his dot Alfred.
From the moment it was activated, Alfred took over every fragmented and complex task running in the background:
- Automated site building and shipping: working from approved design files, Alfred pushed the site code forward on its own, only prompting for confirmation at key decision points (e.g. swapping images #1 and #3) — then published live in seconds.
- Cross-platform tracking and board reporting: it monitored user feedback data in Teams on its own, kept the board deck updated in real time, and pulled the core metrics — usability climbing sharply, annual revenue up 51% YoY.
- Long-horizon architecture refactoring: catching an "app needs migration" note in Google Meet minutes, it traced the whole flow automatically and drafted a complete architecture diagram.

There's also a real internal OpenAI case: an employee forgot to invoice a partner publisher. His dot spotted it, generated the invoice automatically and asked: "Send it now?" One click, and it was handled.
That's the fundamental shift dots represents: it's no longer a tool — it's your digital coworker.
It's like hiring a high-capacity employee who never takes leave, never forgets anything, and is online 24 hours a day. And it keeps learning you — your preferences, your standards, your taste.

Naturally, Altman addressed safety concerns head-on: every dot runs inside an isolated cloud computer, with personal privacy and local data physically isolated by default; when using saved passwords, the plaintext password is never exposed to the model; and when idle it holds only read-only permission for background research — it never sends messages or alters data on its own.
One line from Altman on stage lingers:
"You're not using a tool. You're building a team."
Starting today, Pro, Business Premium and Enterprise users can create their first dot — and those conversations don't consume your base ChatGPT usage allowance!
💼 ChatGPT Space: Workflows, Rebuilt From Scratch
If dots are your digital employees, then ChatGPT Space is the "cloud office" built for those AI employees and their human colleagues.
This past year, every big tech company has been frantically stuffing AI into legacy documents — AI in Google Docs, AI in Notion, Copilot in Microsoft Office.
OpenAI went the other way: it built an "AI-native" collaborative space from the ground up.

Inside Space, you can collaborate seamlessly with colleagues and their dots on the same page:
- All-in-one collaboration: drafting plans, deep research, generating charts, live data analysis — all in one place.
- Real-time interaction:
@Dotin a page the same way you'd@colleague: "turn this data into a bar chart" — the chart appears instantly, with live filtering and drill-down. - Knowledge that sticks: breaking the old pain of chat windows trapping context — teammates and every dot share one knowledge base, so project context never breaks.
In Altman's words: "Pages and files come together in Space — it's like a living cloud drive."
⚡ GPT-6.1 Sol Takes the Stage: Astra-Class Intelligence at 1/5 the Price
Most people assumed there'd be no new flagship at this DevDay — and then Altman pulled out GPT-6.1 Sol to thunderous applause!

Just one week after GPT-6 Sol shipped, OpenAI raised the ceiling again:
On the authoritative software-engineering benchmark DeepSWE v1.1, GPT-6.1 Sol matches the top-tier Astra; on long-horizon business automation AutomationBench and scientific reasoning Terminal-Bench Science, it outright crushes Opus 5.5!
On complex OS operation (OSWorld 2.0), 6.1 Sol trails Astra by just 2.1 points, while its overall cost is roughly 1/7.


And the stunning part: GPT-6.1 Sol shatters the price barrier entirely.
💰 GPT-6.1 Sol vs Astra Pricing
| Model | Input ($/M tokens) | Output ($/M tokens) | Cached context price | Drop vs Astra |
|---|---|---|---|---|
| GPT-6 Astra | $10.00 | $50.00 | Standard | Baseline |
| GPT-6.1 Sol | $2.00 | $10.00 | $0.10 (95% off) | Down 80% (1/5 the price) |

The keynote also introduced Ultrafast inference mode, pushing generation speed to 300 tokens/s (8x standard).
In the live "build a rocket" demo, the Ultrafast rocket had already launched while standard mode was still assembling model parts.

Ultrafast currently prioritizes Astra at launch, then opens up fully to Sol.

To match that call frequency, OpenAI officially revived the $200/month Pro subscription and launched the headline Pro 500 tier ($500/month) — offering 25x the Plus allowance plus native Ultrafast access, removing the ceiling on high-end productivity altogether!

🛠️ Codex Moves to the Cloud: No More Sitting at Your Desk to Ship Code
On the developer front, Codex got an epic upgrade: all of Codex moved to the cloud.
Dev environments can be preconfigured and reused; permissions are set once. Even outdoors, you can control cloud Codex from your phone and keep the project running.

The Codex CLI got a major overhaul — real-time voice commands, mid-flight course corrections, and a new /agents view that tracks multiple concurrent subtasks without chaos.

The ChatGPT desktop app adds automatic code review. While you're away, the AI audits code changes and syncs feedback to GitHub / GitLab.

The new Codex Security Cloud targets vulnerabilities: it scans entire GitHub repositories, dedupes and checks for issues on every commit, and generates fix proposals for review.

The API layer moved just as fast:
- Agents API: the core capabilities that power dots are fully packaged and opened up — computer operation, multi-agent collaboration, tool retrieval, and context compression.
- Decisions API: built for "fast decisions among fixed options" (email triage, UI button choices) with response times as low as 230 milliseconds.

Per Tibo, token traffic on OpenAI's Responses API has grown 100x year over year, while availability held steady at an impressive 99.9%.
⚔️ The Personal Agent War: OpenAI Shows Its Whole Hand
Every one of tonight's 20+ announcements carries weight. And a single logic runs through all of them:
Turn ChatGPT from "a dialog box that answers questions" into "a digital hub that does the work for humans."
Cheaper Sol to cut costs, dots to take over concrete tasks, Space to provide the shared desk, open APIs to plug into the wider ecosystem — OpenAI is moving this fast because the competition has gone white-hot:

- Meta Muse: 3 million downloads in just 3 weeks, powered by an aggressive free strategy;
- Google Spark: already doing 7×24 cloud work on users' behalf;
- SpaceXAI Grok Bot: a strong new entrant shaking up the personal agent track.
Personal agents have, beyond doubt, become the next era-defining gateway that every tech giant is fighting over.
dots' key innovation is a shift in the work paradigm: before you even ask, it has already put a satisfying result in front of you. It no longer asks "what do you need?" — it tells you "it's already done."
Altman closed with:
"People compare AI to a new industrial revolution. I don't love that analogy. If we get this right, it's more like a new Renaissance."


The heart of a Renaissance is freeing humans to spend more time on genuinely creative, meaningful exploration.
When dots takes on all the boring, repetitive, attention-draining trivia — what will you create with the time you get back?
The shift is here: AI is accelerating from "tool" to "colleague."
Your dots is already waiting for you in the cloud!
| Channel | URL | Best for |
|---|---|---|
| High-speed GPT mirror in China | https://trygpt.asia/list/#/home | Browser-based chat without a VPN, lightweight workflow testing |
| MomoAI monthly plans | https://momoai.asia/home | Heavy users who want a monthly subscription |
| All-model API aggregator | https://tryallapi.com/ | Developer calls, low-latency direct connection, one-stop multi-model access |
| Official top-up channel | https://chongzhi.trygpt.asia/ | Removing the overseas payment barrier, official direct top-up |