🔍 Read the full analysis: OpenAI And Anthropic Cut AI Prices While Competing For Subscribers on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI and Anthropic have reduced prices for several AI models in a series of moves since June, while OpenAI is also offering reported discounts to win business customers. Some customers report shifting more work to OpenAI, but headline savings do not describe every deal: usage limits, seat requirements and model performance also affect total cost.
OpenAI and Anthropic have cut prices for several AI models since June, while OpenAI is offering reported discounts to attract business customers from Anthropic. The changes matter to companies and subscribers weighing AI costs, but the figures include different models, contract terms and usage allowances—not just lower prices for the same service.
According to pricing pages and announcements reviewed by ThorstenMeyerAI.com on October 26, OpenAI’s Sol model pricing moved from $5 per million input tokens and $30 per million output tokens in June to $2 and $10 for GPT-6.1 Sol in September. Its smaller Luna model moved from $1 and $6 to 10 cents and 50 cents per million input and output tokens. These are comparisons between successive models and price announcements, not necessarily reductions to an unchanged model.
Anthropic also lowered prices or costs across its model line. Its Opus pricing went from $5/$25 to $4/$20 per million tokens between Opus 5 and Opus 5.5, while cache reads for the newer model were reported at 60 percent less. Anthropic has also said Fable 5.1 can cost around 25 percent less on typical workloads and up to around 45 percent less on highly agentic workloads. Those workload savings are the company’s estimates, not independent measurements.
The customer deals reported by The Information vary. It reported that OpenAI offered Replit a discount of more than 30 percent on a contract worth over $100 million annually, according to a person familiar with the offer. DataCamp CEO Jonathan Cornelissen described expected savings of 80 to 90 percent from OpenAI pricing against his company’s current setup. That figure is not a 90 percent reduction from OpenAI’s list price.
Some customers have also described shifting work between providers. Rippling CEO Parker Conrad said OpenAI’s share of coding work at his company rose from 16 percent in May to 52 percent, while Anthropic’s share fell from 57 percent to 18 percent. He attributed the shift to cost. Replit CEO Amjad Masad said the company had stopped working with Anthropic and called ChatGPT better, but later told TechCrunch that Replit remained model agnostic and OpenAI was its preferred partner. A person familiar with the partnership told TechCrunch it remained in place.
Infographic 01 · The price slide
Four months, a fraction of the price.
What a million output tokens cost on each vendor’s price list, from the June previews to today. Bars show the price; shorter is cheaper.
OpenAI · output price per 1M tokens
List prices, standard tier
Anthropic · output price per 1M tokens
List prices, standard tier
Same scale in both panels. Fable 5.1 kept its $50 output price and cut cache reads instead.
The quieter cut: cache reads
matters most for agents
Cache-read price per 1M tokens, same scale in all three.
Woolf estimates cached tokens make up 96–99% of token spend in agentic coding. Anthropic says Opus 5.5 costs 40% less than Opus 5 and Fable 5.1 about 25% (up to about 45%) less than Fable 5 on typical workloads; those are vendor claims. Most rows compare a successor model with its predecessor. Anthropic’s newer tokenizer produces about 30% more tokens for the same text, so per-token prices are not directly comparable across vendors.
Sources: OpenAI API pricing (26 Oct 2026); OpenAI announcements (26 Jun–29 Sep 2026); Anthropic API pricing (26 Oct 2026); Anthropic announcements (30 Jun–22 Sep 2026); Vercel AI Gateway index (Oct 2026); Max Woolf (25 Oct 2026).
Powered by Thorsten Meyer AI
Lower Prices Reshape AI Contracts
Lower token rates can reduce bills for organizations that send large volumes of requests, particularly when they use coding agents or other systems that repeatedly process conversation history. But the headline price is only one part of the bill: cached-token rates, included usage, seat minimums and contract discounts can change what a customer actually pays.
The reported customer shifts suggest that price and perceived model performance are influencing purchasing decisions. They do not establish that one provider is cheaper or better for every customer. Workloads differ, and a successor model’s performance on a company’s own tasks may not match a vendor’s comparisons. Buyers need to test models against their actual work and compare the full contract, rather than infer savings from a token price alone.
Individual subscribers face a similar distinction. The source says OpenAI reduced usage included in new $200 Pro subscriptions without changing the monthly price. Anthropic also added capped, expiring API credits to some Max and Team plans. A lower model price does not automatically mean more included usage; plan limits can shift even when the fee does not.
Popular right now
Dyson Ball Animal 2 Plus Corded Vacuum
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, Almond Cream
Samsung 48-Inch S90H OLED 4K Smart TV (2026)
As an affiliate, we earn on qualifying purchases.
A Four-Month Pricing Run
Data scientist Max Woolf charted a series of OpenAI and Anthropic pricing moves from June to September 2026. The sequence included OpenAI cutting Luna prices in July, making a temporary Sol reduction in August, and introducing lower prices for newer models in September. Anthropic set introductory Sonnet 5 pricing as permanent, then released Opus 5.5 and Fable 5.1 with different prices or cache costs. ThorstenMeyerAI.com said it checked Woolf’s figures against vendor announcements and pricing pages retrieved October 26.
Comparisons need care because several changes accompanied a new model release. The source notes that whether a successor performs as well on a specific customer’s tasks is something that customer must test. Token prices also do not capture every workload: Woolf estimates cached tokens account for 96 to 99 percent of token spending in agentic coding. That estimate helps explain why cache pricing can matter, though the figure is his assessment.
Contract terms can limit who benefits. Anthropic’s Enterprise plan requires 20 seats, according to the source, a minimum that could put it out of reach for many small advisory firms. The source also reports that OpenAI reduced usage included with new $200 Pro subscriptions, while Anthropic added capped, expiring API credits to Max and Team plans. These terms are separate from advertised per-token rates.
“OpenAI’s share of coding work rose from 16 percent in May to 52 percent, while Anthropic’s fell from 57 percent to 18 percent.”
— Rippling CEO Parker Conrad, as reported by The Information
Actual Savings Depend on Usage
The available reporting does not provide the full terms of every customer contract, including how long discounts last, which models or volumes they cover, and whether usage commitments or other fees apply. The Information’s article is paywalled; the source account says it relied on its public summary and write-ups by The Decoder and all-ai.de for some deal details.
It is also unclear how broadly the reported pricing offers are available. A discount negotiated by a large customer may not be available to smaller buyers. The supplied source does not establish that every reported customer has completed a provider switch or that the changes will persist.
Finally, price comparisons do not settle whether newer models deliver equivalent results for a particular task. Vendor estimates of workload savings and claims about model capability need to be distinguished from independently verified outcomes. Customer-specific performance and total cost remain open questions.
Watch Contracts and Usage Limits
Companies considering a renewal can compare competing offers, test models on representative tasks and review the full contract, including token rates, cached reads, seat minimums, included usage, credit expiry and discount duration. The reported price cuts give buyers a reason to seek updated terms, but the savings should be calculated against their own usage.
Subscribers should check plan allowances as well as monthly fees, especially when plans change without a price adjustment. Further vendor pricing announcements, renewed contracts and customer disclosures may show whether the current cuts lead to lasting reductions in bills or mainly shift spending through different limits and incentives.
Key Questions
How much did OpenAI cut its model prices?
According to the source’s review of vendor pricing, Sol moved from $5/$30 to $2/$10 per million input/output tokens between June and GPT-6.1 Sol in September. Luna moved from $1/$6 to 10 cents/50 cents. These figures compare model and pricing updates; they do not show that an unchanged model received the same reduction.
Did OpenAI offer customers a 90 percent discount?
Not as a general list-price discount, based on the supplied reporting. DataCamp’s CEO described expected total savings of 80 to 90 percent against the company’s existing setup. The Information also reported a discount of more than 30 percent for Replit.
Are customers switching from Anthropic to OpenAI?
Some reported figures show changing usage. Rippling’s CEO said OpenAI’s share of coding work rose from 16 percent in May to 52 percent, while Anthropic’s fell from 57 percent to 18 percent. Replit described OpenAI as its preferred partner, while also saying it remained model agnostic.
Why do usage limits matter if prices are falling?
A monthly plan or token rate does not show how much a customer can use before extra costs or limits apply. The source reports changes to OpenAI Pro usage allowances and Anthropic API credits, along with an Enterprise seat minimum. Buyers should check those terms alongside the advertised price.
Source: ThorstenMeyerAI.com
