🔍 Read the full analysis: The 5X Explained: What SemiAnalysis Found In AI Subscriptions on ThorstenMeyerAI.com
Get everyday essentials delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
TL;DR
SemiAnalysis measured usage limits across major AI subscriptions and compared them with the providers’ API list prices. In its comparison of agentic workloads, Claude’s mid-tier plans offered about 5.4–5.6 times the API-equivalent value of comparable ChatGPT plans, but recent limit and price changes have already shifted the calculations. The report also estimates that subscriptions use a much larger share of Anthropic’s inference compute than of its revenue.
SemiAnalysis has compared usage allowances across major AI subscriptions by measuring how their limits move as different types of tokens are used, then pricing that usage at each provider’s API list rates. For an agentic workload, its analysis estimates that Claude’s mid-tier plans provide about 5.4–5.6 times the API-equivalent value of comparable ChatGPT plans, a finding that comes as both companies adjust prices and limits.
The report’s central comparison covers three price points: $20, $100 and $200 per month. At those tiers, SemiAnalysis estimates API-equivalent usage of $1,178 for Claude Pro, $5,725 for Claude Max 5x and $11,726 for Claude Max 20x. Comparable ChatGPT plans return estimates of $211, $1,055 and $2,084, respectively. The ratios range from about 5.4 to 5.6 times in Claude’s favor.
Those figures use a coding-agent workload dominated by cached input: SemiAnalysis gives its mix as roughly 96.6% cached input, 2.6% cache writes, 0.4% fresh input and 0.3% output. The report defines “API value” as the plan’s full monthly usage limit priced at first-party API list rates. Because GPT-6.1 Sol is cheaper per token than Claude Opus 5.5, the dollar comparison favors the more expensive model; SemiAnalysis says the gap also remains large when measured in raw tokens.
The frontier-model comparison is closer. SemiAnalysis says the allowances for GPT-6 Astra and Claude Fable 5.1 are broadly similar. On a $200 plan, it estimates that OpenAI’s allowance would cover about $2,897 of Astra usage, while Fable would use about half of a Claude plan’s limit at $2,485. That remaining capacity can be used for other Claude models, including Opus or Sonnet, which affects the overall value comparison.
The 5x is a subsidy, not a price
SemiAnalysis metered the meters — every major AI subscription, token type by token type, converted to API list value. On the mid-tier models both labs call the daily driver, a Claude plan returns ~5–6× the API value of the matching ChatGPT plan. Real — and the least durable number in the report.
…and the plan is fully exhausted. One pool for every model.
…and the plan is only half used — Fable is capped at 50% of the limit, leaving the rest for Opus/Sonnet. That’s where the mid-tier gap compounds.
- $200 plan halved — Sol-class value down >50% (6.1 Sol cache price cut compounds it)
- Old limits kept until 29 October; new buyers cut immediately
- New $500 tier: only +21% Astra vs the old $200 — real draw is 300 TPS Ultrafast
- Ladder flattened: Pro 100/200/500 now identical per dollar; multipliers removed from pricing page
- In OpenAI’s favour: no 5-hour window on Pro plans — easier to use the full allowance
- Flat per-dollar value across all tiers, before and after
- New premium models placed at lower relative limits (Fable capped at 50%)
- Opus allowances raised ~20% (Max) / ~50% (Pro) with the 5.5 price cut — not enough to fully offset it
- Repeatedly walked back planned cuts earlier this year under pressure from OpenAI’s generosity
- Twelve months ago, OpenAI was the generous option. Positions swap.
Gross margin per plan, assuming 92% API gross margins. The subsidy lives almost entirely in Opus and Sonnet usage — Anthropic would already be near software-like subscription margins if everyone used only Fable. Subscriptions matter even more for OpenAI, where they’re a larger share of revenue.
Three identical subscriptions; one had ~20% lower limits. The provider (unnamed) confirmed an “extremely tiny” A/B test on limit balancing. Two lessons: limits can change silently, per account, at any time — and you won’t know without instrumentation. The usage bar is a percentage, not a contract.
If you’re choosing a plan this month for agentic coding on a mid-tier model, the report settles it: a Claude plan returns ~5–6× the API value of the matching ChatGPT plan. But a plan returning 58× its fee on a model served at a steeply negative margin for heavy users is a marketing budget with a usage meter. Value moves silently, gets A/B tested per account, and twelve months ago ran the other way. Use the subsidy while it exists — it’s genuinely large. Don’t build a cost model on it. Price workloads at API rates, keep a router between you and any one vendor, and benchmark open weights on your own hardware for steady volume. A deal you can’t verify isn’t a price. It’s weather.
Subscription Value Meets Compute Costs
The value gap matters because subscriptions affect providers’ costs as well as their ability to attract and retain customers. SemiAnalysis estimates that subscriptions represent about 10% of Anthropic’s revenue but can consume more than 40% of its inference compute. The report estimates this mix lowers Anthropic’s blended revenue per megawatt by roughly $36 million. It says subscriptions account for a larger share of OpenAI’s revenue, though the supplied material gives no corresponding compute estimate for OpenAI.
SemiAnalysis models the economics of subscribers who use all of their allowance, assuming API gross margins of 92%. Under those assumptions, it estimates gross margins of about minus 369% for fully used Opus 5.5 subscriptions and about 1% for Fable 5.1. At 20% average utilization, its estimates rise to roughly 6% for Opus and 80% for Fable. These are modelled outcomes, not reported financial results, and depend on the stated utilization and margin assumptions.
The analysis suggests that the generosity of a subscription can be costly when customers rely heavily on premium models. At the same time, actual costs depend on how much subscribers use their allowances and which models they choose. SemiAnalysis’s estimates help explain why providers may alter limits even when a monthly fee stays the same.
As an affiliate, we earn on qualifying purchases.
Recent Price and Limit Changes
OpenAI’s recent changes are central to the comparison. SemiAnalysis says OpenAI halved the allowances on its $200 plan, with new purchases subject to the lower limits immediately. Existing subscribers keep the former limits until October 29. The source material does not specify the year for that date. The report says the value of Sol-class usage fell by more than half because OpenAI also lowered GPT-6.1 Sol’s cached-input price.
OpenAI introduced a $500 tier as part of the changes. SemiAnalysis estimates it offers about 21% more Astra usage than the former $200 plan, but less Sol-class API value, in part because of the Sol price reduction. The report identifies 300 tokens per second “Ultrafast” mode as the tier’s main distinction and says it was still testing that feature. It also says OpenAI removed “5x more usage” and “20x more usage” comparisons from its pricing page.
Anthropic has also cut model prices. According to SemiAnalysis, Fable 5.1 reduced cache-read prices by 75% compared with Fable 5, while Opus 5.5 cut input and output prices by 20% and cache reads by 60% compared with Opus 5. The report says Fable’s limits did not rise at launch; Opus allowances increased by about 20% on Max and 50% on Pro, which did not fully offset the price reductions. It also says OpenAI did not raise Sol limits when GPT-6.1 shipped.
As an affiliate, we earn on qualifying purchases.
How Long Limits Will Hold
The reported ratios describe a measured workload and a particular set of plan limits and API prices. They are not a guarantee of the value every subscriber will receive: usage mix, model choice and how much of an allowance a person uses can change the result. SemiAnalysis says OpenAI’s Pro plans do not have a five-hour usage window, which may let heavy users spend more of their monthly allowance in practice, especially during bursts. The report judges that this does not erase the estimated value gap, but the practical effect will vary by workload.
The figures may also change as providers revise model prices, plan limits or subscription terms. The supplied material does not give the full test protocol, the measurement dates for every plan, or a detailed uncertainty range for the estimates. It says SemiAnalysis was still testing OpenAI’s Ultrafast mode. The report’s estimates of compute use and subscription margins are based on assumptions; they are not audited figures from either company.
As an affiliate, we earn on qualifying purchases.
Plan Changes Set the Next Test
The next clear point for current ChatGPT Pro customers is October 29, when the reported extension of the old $200 plan limits for existing subscribers ends. The supplied material does not say what limits those subscribers will receive afterward or whether OpenAI will revise the terms again. New purchasers, according to the source, already receive the reduced allowance.
Further changes to API prices and subscription limits will affect the comparison, particularly if providers raise allowances after cutting model prices. SemiAnalysis’s figures provide a snapshot of the plans it measured; readers should check current plan terms before treating the reported ratios as a present-day estimate for their own use. The report says its testing of Ultrafast mode is ongoing, but the supplied material gives no publication date for those results.
As an affiliate, we earn on qualifying purchases.
Key Questions
What did SemiAnalysis measure?
It measured how usage limits on major AI subscriptions respond to different token types, then priced the estimated usage at the providers’ API list rates.
What does the 5.4–5.6× figure compare?
It compares the estimated API-equivalent value of Claude and ChatGPT plans at $20, $100 and $200 per month for the agentic workload described in the report. It does not mean every user will receive that ratio in practice.
Why did OpenAI’s reported plan value fall?
SemiAnalysis says OpenAI roughly halved allowances on the $200 plan and cut the cached-input price for GPT-6.1 Sol. The lower API price reduced the dollar value assigned to a given amount of Sol usage.
Did Anthropic raise every model’s subscription limit after price cuts?
No. SemiAnalysis says Fable 5.1 launched without higher token limits. It says Opus allowances rose by about 20% on Max and 50% on Pro, increases that did not fully offset the model’s price reductions.
Are the subscription margin figures company-reported results?
No. They are SemiAnalysis estimates based on stated assumptions, including 92% API gross margins and different levels of subscriber utilization.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
