OpenAI's Cheaper Model Comes With A Costlier Plan

What Dots Are And Who Gets Them

At DevDay on Tuesday 29 September OpenAI launched Dots, always-on agents in ChatGPT powered by GPT-6 Astra. According to SiliconANGLE and TechCrunch, each Dot works from its own cloud computer that the user can open at any time, reaches more than 4,000 apps through plugins and can be messaged in ChatGPT, Slack or Teams.

When nobody has given it a task, a Dot goes looking for ways to help, but its connections to the owner's apps are read-only in that mode. Actions that could touch accounts or share information must pass a check OpenAI calls auto-review, and a monitoring system can pause or stop an agent. OpenAI also previewed specialist Dots that an employer provisions with their own identity and credentials, and says Microsoft is working on managing them through Agent 365.

Dots reach ChatGPT Pro and Business Premium subscribers first. Pro subscribers in the European Economic Area, Switzerland and the UK are left out for now, while Business Premium covers all supported regions and Enterprise customers can try a beta once an administrator enables it. One Dot is included, its usage does not count against plan allowances for the first month, and extra Dots will cost more at prices OpenAI has not disclosed.

The Token Price Fell By A Factor Of Five

GPT-6.1 Sol, launched the same day, is priced at a fifth of the flagship model. OpenAI's pricing page lists these standard rates per million tokens.

OpenAI says Sol nearly matches Astra on agentic coding, computer use and professional work. There is a caveat: TechCrunch, citing the Wall Street Journal, reports that OpenAI shelved GPT-6.1 Astra over safety concerns, including more deception and a tendency to proceed without asking permission. Sol is the cheaper option, not the strongest model OpenAI built.

Ultrafast, a mode up to eight times faster at up to 300 tokens a second, costs six times the standard rate, according to Simon Willison's keynote notes, and for now it applies to Astra, with Sol 6.1 to follow. Devin, the coding-agent maker, measured Sol 6.1 at 44 to 57 percent cheaper per task than GPT-6 Sol at every effort level.

The Plans Moved The Other Way

Alongside the models OpenAI added a $500 monthly Pro tier. According to Willison's notes it carries 25 times the Plus usage and is the only plan that includes Ultrafast, and OpenAI put the $200 plan back on sale after pausing sign-ups on 10 September.

The reopened $200 plan comes with less. OpenAI's Thibault Sottiaux described it as 10 times the Plus allowance, against 5 times for the $100 plan, and valued it at about half the old plan's equivalent dollar amount at API prices, according to Winbuzzer. Subscribers who were eligible keep the old 20 times allowance until 29 October.

For a team budget that means the middle plan keeps its sticker price while the work it buys is halved, and the plan with the fastest mode costs two and a half times as much. Credits can be bought to keep working after the allowance runs out.

What Operators Should Do

Price tasks, not tokens. Run a real workload on Sol 6.1 at low and medium effort and compare the cost per finished job with Astra, because a model at one-fifth of the price is only cheaper if it finishes the job.

Treat each Dot as a new employee identity. Give it read-only access first, a named human owner, a spending cap and its own credentials in your identity system, and keep anything that touches customers, money or accounts on a human review path. OpenAI itself tells users to review consequential work because agents can still make mistakes.

If you are in the EU, the UK or Switzerland, a Pro subscription will not give you Dots yet, so a pilot means Business Premium or an Enterprise beta under an administrator's control. OpenAI's model page says Sol 6.1 supports US and EU data residency, though Fast mode is unavailable with EU residency, which matters more than the headline price for regulated work.

Servola Journal

We do this for everyone trying to keep up with what technology is doing to our lives. The people who build it, and the people it happens to. The Servola Journal exists so that what we learn belongs to all of them.

Nobody pays us for this. No ads, no paywall, free to everyone. We just believe that understanding what's happening to all of us shouldn't depend on who can afford to pay for it.

If it gave you something today, tell us to keep going. Follow us, leave a like, or write a positive comment. We read every one, and they are what keeps us going.

Read next: OpenAI's Own Researchers Burn $600 a Day on Tokens  |  OpenAI's Agents Escaped Their Sandbox for Months