OpenAI rewrote the price ladder for paid ChatGPT plans on the same day it unveiled its newest products, and the people who pay the most noticed first. According to AINews, the daily digest published by Latent Space, the new lineup runs Plus at 1x, Pro 100 at 5x, Pro 200 at 10x, and a new Pro 500 at 25x. The digest says the change roughly halves what the old Pro 200 tier was worth, and that it drew heavy backlash.

The multipliers measure how much usage each plan allows compared with Plus. The digest does not print monthly prices, but if the tier names denote dollars per month, every step costs about $20 per unit of usage: $100 for 5x, $200 for 10x, $500 for 25x. That would make the ladder perfectly linear, and it would mean heavy users who once paid $200 for a larger allowance now get less for the same money unless they move up to the $500 rung. This is an inference from the labels, not a figure OpenAI has confirmed in the material AINews summarized.

The reaction showed up on X. The digest’s list of the most-engaged posts of the day includes one from Tibo on the Pro $200 usage recalculation (27.9K engagement) and another on the new plan multipliers (12.3K). The first out-drew the posts announcing Sol (20.8K) and Sam Altman’s note that Dots had arrived (13.1K); only OpenAI’s own Dots announcement did better. On that measure, the quota changes drew more attention than most of the launches themselves.

Those launches were substantial. OpenAI introduced Dots, always-on agents that each run on their own cloud computer, and it says they ship to Pro, Business Premium, and Enterprise customers. According to the digest, Tibo clarified that a primary dot’s own work does not draw down plan usage, while any Codex tasks it spawns do. That split matters for the new tiers, because the agents OpenAI is promoting are the ones most likely to burn through a quota.

On the developer side, OpenAI priced GPT-6.1 Sol at $2 per million input tokens and $10 per million output tokens, and a cache hit costs only $0.10, which the digest calls a 95 percent discount. OpenAI pitches Sol as near the intelligence of its flagship Astra at a fifth of the price. The company also announced an Ultrafast mode that it says generates up to 8x faster in Codex, at 300 tokens per second, and 6x faster in the API. Speed carries a premium: Ultrafast is priced at 6x, which works out to $60 and $300 per million tokens for Astra.

Independent numbers so far are mixed but favorable on cost. Artificial Analysis, an independent benchmarking firm, placed Sol one point below Astra on its Intelligence Index, at $0.72 per task against $3.26, according to the digest. Separately, one engineer, identified in the digest by an X account, seeded two code repositories with 105 bugs. Sol caught 44 of them for $6.56, and Astra caught 45 for $33. Those are single tests by individuals, and the digest notes that a harness choice moved results in other runs.

The pattern across the day is a company steering heavy users toward pay-as-you-go economics. Cheap, capable Sol tokens and a metered speed tier reward developers who count cost per task, while the subscription ladder now charges a steeper rate for the biggest flat-fee allowances. OpenAI has not published the arithmetic behind the recalculation in anything AINews cited.

Anyone who built a workflow around a Pro 200 seat should measure actual weekly usage this week and price it against Sol on the API, because at $0.72 per task the metered route may now beat the $500 rung for steady agent workloads.

Reported by AINews on Latent Space on 30 September 2026.