🔍 Read the full analysis: A Price Tag Or A Subsidy? SemiAnalysis On The 5X In AI on ThorstenMeyerAI.com
Get smart everyday buys 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 compared usage limits across major AI subscriptions by measuring token consumption and valuing it at API list prices. Its agentic-workload comparison found Claude mid-tier plans offered roughly five to six times the API-equivalent usage of comparable ChatGPT plans, but recent price and allowance changes make the gap subject to change. The report also estimates that heavy use of premium models can make subscriptions costly for providers.
SemiAnalysis has published a comparison of AI subscription usage limits across major providers, estimating that Claude plans deliver about five to six times the API-equivalent usage of comparable ChatGPT plans on a coding-agent workload. The finding offers a measure of how much usage subscribers may receive for a fixed monthly fee, while the report says recent price and allowance changes make the comparison time-sensitive.
To make the comparison, SemiAnalysis measured how each provider’s usage bar moved across token types, then priced the measured usage at first-party API list rates. Its central comparison uses an agentic workload dominated by cached input: about 96.6% cached input, with roughly 0.4% fresh input, 2.6% cache writes and 0.3% output. The report calls the resulting dollar figure “API value”; it represents the plan’s full monthly limit priced as API usage, not cash paid out to subscribers.
For the selected mid-tier models, SemiAnalysis estimates that a $20 Claude Pro plan allows about $1,178 of API-priced Opus 5.5 usage, compared with $211 of GPT-6.1 Sol usage on ChatGPT Plus, a ratio of about 5.6 to 1. At the $100 and $200 tiers, the estimated ratios are about 5.4 and 5.6 to 1. SemiAnalysis says the gap remains large when comparing raw token quantities, though the models’ different prices affect the dollar valuations.
The report also describes recent changes at both companies. It says OpenAI roughly halved usage allowances on its $200 plan, with existing subscribers keeping prior limits until October 29 while new purchases receive the reduced limits immediately. It says Anthropic lowered API prices for newer models, while changes to subscription allowances varied by model and tier. Those details affect what a plan buys, so the figures are a snapshot rather than a fixed measure of subscription value.
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 measured gap matters to subscribers comparing monthly plans, but the report’s larger point is that a high allowance can carry a substantial cost for the provider. SemiAnalysis estimates subscriptions account for about 10% of Anthropic revenue while consuming more than 40% of its inference compute. It says that mix lowers blended revenue per megawatt by roughly $36 million, based on its estimates.
The report models gross margins under different usage assumptions. Assuming a subscriber fully uses the plan and API gross margins are 92%, it estimates Opus 5.5 usage could imply a gross margin near minus 369% for the subscription, while Fable 5.1 usage would imply about 1%. At 20% average utilization, its estimates rise to about 6% for Opus and 80% for Fable. These are modeled outcomes, not reported company results, and depend on the assumptions in the analysis.
That distinction helps explain why a plan can look unusually generous and still be commercially useful: many subscribers may not exhaust their limits, and providers can adjust allowances as models and costs change. SemiAnalysis says subscriptions make up a larger share of OpenAI revenue than Anthropic revenue, though the source material does not give a precise OpenAI share.
As an affiliate, we earn on qualifying purchases.
Recent Model Prices and Plan Limits
The comparison follows several model and pricing changes described by SemiAnalysis. It says Anthropic’s Fable 5.1 cut cache-read prices by 75% against Fable 5, while Opus 5.5 cut input and output prices by 20% and cache reads by 60% against Opus 5. The report says Fable 5.1 launched without higher token limits; Opus allowances rose about 20% on Max and about 50% on Pro, still short of fully offsetting the reported price reductions.
For OpenAI, the report says GPT-6.1 Sol’s cached-input price fell without a corresponding change in its limits, reducing the API-equivalent value of the $200 plan by roughly 30%. It also says the company introduced a $500 tier, with an Ultrafast mode advertised at 300 tokens per second. SemiAnalysis was still testing that mode when it prepared its analysis.
The source says OpenAI removed “5x more usage” and “20x more usage” comparisons from its pricing page. It also notes that ChatGPT Pro plans do not have a five-hour usage window, which may let heavy users spend more of their monthly allowance in practice. On frontier models, the report describes a closer comparison: it says a $200 plan’s allowance is exhausted after roughly $2,897 of GPT-6 Astra usage, while Claude’s allowance is half-used after roughly $2,485 of Fable 5.1 usage. The report attributes that split to Fable using only half of the Claude plan’s limit.
As an affiliate, we earn on qualifying purchases.
Limits Behind the Value Estimates
The estimates depend on measured usage-bar movement, selected model prices and the chosen coding-agent token mix. Actual subscriber usage varies, and the report’s full-utilization margin scenarios do not describe typical customer behavior. The source material does not provide the complete measurement protocol or uncertainty ranges for each plan.
It is also unclear from the source how long the reported limits will remain in place. OpenAI’s grandfathering date applies to existing $200 subscribers, while the report describes other allowance and price changes that could alter the ratios. SemiAnalysis had not finished testing Ultrafast mode, and the material provides no later results or confirmation of its performance in ordinary use.
As an affiliate, we earn on qualifying purchases.
Watch for Further Limit Changes
The next clear date identified in the report is October 29, when existing $200 ChatGPT subscribers are due to lose their previous limits under OpenAI’s stated change. Subscribers and plan comparisons will also be affected if either company changes model prices, monthly allowances, or the models available within a tier.
SemiAnalysis said it was still testing OpenAI’s Ultrafast mode. Further measurements could clarify whether the new $500 plan’s speed feature changes its practical value, alongside its stated usage limits. Until those results and any subsequent plan updates are available, the five-to-six-times estimate should be read as a comparison of the specified plans, models and workload at the time measured.
As an affiliate, we earn on qualifying purchases.
Key Questions
What does the reported five-to-six-times gap measure?
It compares estimated usage limits for selected Claude and ChatGPT plans on a coding-agent workload, with measured usage priced at each provider’s API list rates. SemiAnalysis estimates the Claude plans deliver roughly five to six times the API-equivalent value on the mid-tier models it tested.
Does API-equivalent value mean subscribers receive that amount of money?
No. It is an estimate of what the measured plan allowance would cost at API list prices. It is not a cash payment, refund or guarantee that a subscriber can use the entire allowance.
What changed for OpenAI’s $200 plan?
SemiAnalysis says OpenAI roughly halved its token allowances. According to the source material, new purchases receive the lower limits immediately, while existing $200 subscribers keep their prior limits until October 29.
Why could a subscription be costly for an AI provider?
Heavy use of premium models can consume substantial inference compute. SemiAnalysis models very low or negative gross margins under full-utilization assumptions for some plans, but its estimates change substantially when it assumes lower average usage.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
