The Information posted on X on 4 November that OpenAI and Anthropic “are releasing cheaper AI models that can sharply reduce customers’ bills, potentially undermining sales of their more expensive offerings”. In each lab’s main current lineup, as listed on its own pricing pages, the cheapest model costs $0.10 per million input tokens and $0.50 per million output tokens, against $10 and $50 for the most expensive: 1% of the top price, by our arithmetic.
The post links to a 2 November AI Agenda newsletter by Stephanie Palazzolo, It’s Contract Renewal Season, But Some Customers Aren’t Biting. Its free opening reads: “My colleagues and I have been hearing a lot of stories about big enterprise customers renegotiating their annual contracts with model developers”. The rest is behind a paywall; we have not read it.
Business Pill · The Jevons Paradox Explained
A short explainer on why cheaper, more efficient inputs can raise total use. It teaches the general idea only and does not discuss OpenAI, Anthropic or their prices.
Output price per million tokens at standard rates, each lab’s main current lineup ranked by list price: $50 at the top and $0.50 at the bottom for both (Haiku 5.5 for prompts up to 100,000 tokens). Source: OpenAI and Anthropic API pricing pages, read 4 November 2026; grouping and ratios are our arithmetic. Chart: FourWeekMBA.
The key insight: As we read it, the two labs now price their main lineups on the same 100-to-1 ladder: $10 and $50 per million tokens at the top, $0.10 and $0.50 at the bottom, so 1 billion output tokens cost $50,000 or $500 depending on the rung, by our arithmetic.
OpenAI’s API pricing page lists four models under the heading Flagship models, at standard processing rates. GPT-6 Astra costs $10 per million input tokens and $50 per million output tokens; GPT-6.1 Sol $2 and $10; GPT-6.1 Terra $1 and $5; and GPT-6 Luna $0.10 and $0.50.
Anthropic’s pricing page lists the four current models on Anthropic’s models overview: Claude Fable 5.1 at $10 and $50, Claude Opus 5.5 at $4 and $20, Claude Sonnet 5.5 at $2 and $10, and Claude Haiku 5.5 at $0.10 and $0.50 for prompts up to 100,000 tokens. Haiku 5.5 prompts over 100,000 tokens cost $0.50 and $2.50.
Within these two main lineups, the top and bottom rungs carry the same list prices. Both pages also list pricier models outside them: OpenAI’s cyber model GPT-5.6 Cyber at $12.50 and $75, and Anthropic’s older Claude Opus 4.1 and Opus 4 at $15 and $75. Sol sits at a fifth of Astra’s price and Terra at a tenth; Opus 5.5 is 40% of Fable 5.1 and Sonnet 5.5 a fifth, by our arithmetic.
Output price per million tokens at standard rates: GPT-6 Astra and Claude Fable 5.1 $50, Claude Opus 5.5 $20, GPT-6.1 Sol and Claude Sonnet 5.5 $10, GPT-6.1 Terra $5, GPT-6 Luna and Claude Haiku 5.5 $0.50 (Haiku 5.5 for prompts up to 100,000 tokens). Source: OpenAI and Anthropic pricing pages, read 4 November 2026; the 1% ratio is our arithmetic.
What Was Posted on X
OpenAI and Anthropic are releasing cheaper AI models that can sharply reduce customers’ bills, potentially undermining sales of their more expensive offerings.
The Information posted on X on 4 November that OpenAI and Anthropic are releasing cheaper AI models that can sharply reduce customers’ bills, potentially undermining sales of their more expensive offerings. This piece checks that against the labs’ own price lists.
What the Cut Looks Like on a Bill
Take a workload that produces 1 billion output tokens. At list prices, by our arithmetic, it costs $50,000 on GPT-6 Astra or Claude Fable 5.1, $10,000 on GPT-6.1 Sol or Claude Sonnet 5.5, and $500 on GPT-6 Luna or Claude Haiku 5.5.
Two of Anthropic’s newer models are priced below the ones they follow, while Fable 5.1 kept Claude Fable 5’s $10 and $50. Opus 5.5 lists at $4 and $20 against $5 and $25 for Claude Opus 5, 20% lower by our arithmetic. Haiku 5.5, at $0.10 and $0.50, is a tenth of Claude Haiku 4.5’s $1 and $5, by our arithmetic. Anthropic’s page also says a cache hit on Opus 5.5 and Sonnet 5.5 costs 5% of the standard input price, against 10% on most models.
OpenAI’s cuts came earlier: Luna’s output price fell from $6 per million tokens in July to $0.50, a 92% drop, and Sol’s from $30 to $10, as we reported on 2 November from OpenAI’s changelog.
Why the Per-Token Price Is Not the Whole Bill
A token is not the same unit at both labs. Anthropic’s page says Claude 4.7 and later models use a newer tokenizer that “produces approximately 30% more tokens for the same text”. The ratios inside each lab’s ladder hold; a cross-lab comparison at equal list prices does not mean equal cost for the same text.
Cheaper models are also not drop-in replacements. Anthropic describes Haiku 5.5 as “For high-volume, latency-sensitive tasks such as classification, extraction, and routing”, and Fable 5.1 as “For demanding reasoning and long-horizon agentic work”. The saving applies to the work a customer can move down the ladder.
Connecting the Coverage
Price pressure also comes from outside the two labs: Kimi K3.1 Pro’s open weights lead Opus 5.5 on 16 of 37 benchmark rows in Moonshot’s own table, and a GLM-based coding agent passed Fable 5 on one benchmark at about 30 times lower cost per trial, as its authors reported.
On the cost side, Elon Musk said Grok Bot’s serving cost was “unsustainably high relative to revenue”. Anthropic’s amended IPO filing shows an operating loss of $1.26 billion for the first nine months of 2026, as we reported. OpenAI’s next models could come as soon as Friday, The Information reported.
The Structural Read
As we read the two price pages, the middle rungs differ more than the ends: OpenAI’s second model, GPT-6.1 Sol, is a fifth of its top price, while Anthropic’s second, Opus 5.5, is 40% of Fable 5.1, by our arithmetic.
As we read The Information’s post, the risk it names is mix, not price alone: a customer that moves work from the top rung to a lower one pays the lower rung’s price for that work.
The Information, on X, 4 November 2026
“OpenAI and Anthropic are releasing cheaper AI models that can sharply reduce customers’ bills, potentially undermining sales of their more expensive offerings”
Three Implications
FOR A 1-BILLION-TOKEN OUTPUT BILL At list prices it is $50,000 on GPT-6 Astra or Claude Fable 5.1 and $500 on GPT-6 Luna or Claude Haiku 5.5, by our arithmetic.
FOR ANTHROPIC’S OWN LINEUP Opus 5.5 lists 20% below Claude Opus 5 and Haiku 5.5 at a tenth of Claude Haiku 4.5, by our arithmetic from its pricing page.
FOR CROSS-LAB COMPARISONS Anthropic says its newer tokenizer produces about 30% more tokens for the same text, so equal list prices do not mean equal cost per text.
The framework says of the token: “It is simultaneously the unit of cognition the model produces, the unit of compute the data center serves, the unit of price the lab charges, and the unit of value the enterprise extracts.”
As we read it, the two price ladders put a 100-to-1 range on the third of those roles, the unit of price, inside each lab’s main lineup. The framework does not discuss these price lists or The Information’s report.
We have read The Information’s X post and the free opening of its newsletter, not the full text. Which customers renegotiated, how much they cut, and whether cheaper models have reduced either lab’s sales are not established here; The Information’s own wording is “potentially”. Neither lab publishes revenue by model. Per-task cost depends on how many tokens each model uses for the same job, which list prices do not show.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
The Information reports that OpenAI’s and Anthropic’s cheaper models can sharply cut customers’ bills. Their own price lists show how far: in each lab’s main current lineup the cheapest model is $0.10 and $0.50 per million tokens against $10 and $50 at the top, 1% of the price by our arithmetic, so 1 billion output tokens cost $500 instead of $50,000.
95,000+ executives read Business Engineer for the AI strategy frameworks cited by ChatGPT, Claude, and Perplexity.
A note on sourcing. The Information’s report is from its X post of 4 November and the visible title, byline and opening of its 2 November AI Agenda newsletter; the rest of the newsletter is paywalled and was not read. Model prices are from OpenAI’s and Anthropic’s API pricing pages, read on 4 November 2026, at standard processing rates. Figures attributed to earlier pieces come from those pieces. Ratios and bill examples are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
