Considering OpenAI Or Anthropic? Check The Terms Behind The Price

🔍 Read the full analysis: Considering OpenAI Or Anthropic? Check The Terms Behind The Price on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get smart everyday buys delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

OpenAI and Anthropic have cut prices on several AI models since June, and OpenAI is also offering discounts to pursue large business customers, according to reporting by The Information summarized by ThorstenMeyerAI.com. The changes give organizations more leverage to negotiate, but headline token prices do not capture the full cost: usage limits, credits, seat requirements and model performance can change the value of a deal.

Thorsten Meyer’s October 26 review of vendor pricing pages and announcements reports that OpenAI’s Sol model pricing moved from $5 input and $30 output per million 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 tokens. Anthropic’s Opus pricing moved from $5 and $25 for Opus 5 in July to $4 and $20 for Opus 5.5; Anthropic also cut cache-read pricing for some models.

Those figures compare different model releases, not necessarily the same model becoming cheaper. Meyer cites data scientist Max Woolf’s October 25 pricing timeline and says he checked the figures against vendor announcements and current pricing pages retrieved October 26. Whether a newer model performs as well as, or better than, its predecessor on a particular company’s work requires testing. Anthropic’s estimates that some newer models cost less on typical or agent-heavy workloads are the company’s claims, not independent guarantees of savings.

The Information reported October 24 that OpenAI is using discounts to attract customers associated with Anthropic. The examples reported include an offer of more than 30% off a contract valued at over $100 million a year for Replit, and a US bank receiving seats for less than $7 each. DataCamp CEO Jonathan Cornelissen described an expected 80% to 90% reduction in his company’s AI costs under OpenAI’s offer. That is an estimate of total savings against its existing setup, not a 90% discount from OpenAI’s list price.

Why Contract Terms Change the Price

Falling list prices make it easier to compare vendors, but the invoice a customer receives depends on more than the posted price per token. A company may pay for subscriptions, API usage, seats, or additional capacity under separate rules. Discounts can also be specific to a customer or contract, so the reported terms do not establish what another buyer would be offered.

Usage policies can change without a monthly fee changing. Meyer reports that OpenAI reduced the usage included in new $200 Pro subscriptions, while Anthropic added capped, expiring API credits to its Max and Team plans. Anthropic’s Enterprise plan has a 20-seat minimum, a threshold that may put it out of reach for smaller firms. Buyers should compare the actual limits and renewal terms with their expected use, rather than treating the monthly fee as the whole price.

Model choice also affects the comparison. If a replacement model behaves differently on a company’s coding, research or support tasks, lower token rates may not translate directly into lower operating costs. For organizations with high usage, the ability to route work between providers may help them compare performance and negotiate, but it also adds operational work. Real-world testing and contract review are needed to establish whether a price cut produces savings for a particular customer.

A Four-Month Price-Cut Sequence

The reported reductions followed a series of launches and pricing changes from June through September 2026. Anthropic introduced Sonnet 5 at introductory pricing in June and made that pricing permanent in August, according to Meyer’s timeline. OpenAI cut Luna pricing in July and reduced Sol pricing in August; September brought new OpenAI models at lower posted rates and Anthropic’s Opus 5.5.

Some cost changes concern cached tokens, which are repeated inputs that can receive a lower rate, rather than the standard input and output rates shown in a headline comparison. Woolf estimates that cached tokens account for 96% to 99% of token spending in agentic coding workloads. That estimate helps explain why cache pricing may matter to some developers, but actual costs depend on how a customer’s systems use the models.

The customer reports also describe movement between providers. Rippling CEO Parker Conrad said OpenAI’s share of the company’s coding work rose from 16% in May to 52%, while Anthropic’s share fell from 57% to 18%. Replit CEO Amjad Masad said at TechCrunch Disrupt that the company was no longer working with Anthropic and that he thought ChatGPT was better; he later told TechCrunch that Replit remained model-agnostic and OpenAI was its preferred partner. A person familiar with the partnership told TechCrunch it was still in place.

“OpenAI’s share of the company’s coding work rose from 16% in May to 52%, while Anthropic’s fell from 57% to 18%.”

— Rippling CEO Parker Conrad, as reported by ThorstenMeyerAI.com

What Buyers Still Need to Verify

The available reports do not provide the full terms of every customer offer, including all usage caps, contract duration, renewal provisions or eligibility requirements. The Information’s article is paywalled; Meyer says his account draws on its public summary and write-ups by The Decoder and all-ai.de as well as other reporting. Deal-specific details should be treated as reported accounts, not as standard prices available to every customer.

It is also unclear whether every price reduction will remain in effect. One OpenAI reduction described in Meyer’s timeline was announced for a three-month period, and vendor pricing and subscription allowances can change. The reported comparisons also involve different model versions and workload assumptions, so they do not establish a universal ranking of cost or capability. Customers would need their own usage data and tests to calculate a reliable comparison.

Compare Offers Against Actual Use

Companies considering a renewal or provider change can use the recent cuts as a reason to request updated pricing and written terms. A useful comparison should include input, output and cache rates, expected monthly usage, included allowances, credit expiry dates, seat minimums, overage charges and renewal conditions. Buyers should test candidate models on their own tasks and measure both quality and cost before shifting substantial workloads.

For individual subscribers, the practical next step is to check the current usage allowance as well as the monthly charge. For enterprise buyers, it is to ask vendors to specify which discounts apply to the full contract and which apply only for a limited period. Further changes may follow as vendors compete for customers, but the source material does not establish what future prices or terms will be.

Key Questions

Did OpenAI cut its list prices by 90%?

No. The 80% to 90% figure refers to DataCamp CEO Jonathan Cornelissen’s estimate of total savings against his company’s existing setup under an OpenAI offer. It is not a 90% reduction from OpenAI’s list price.

How much did the cited OpenAI model prices change?

According to the pricing figures reviewed by Thorsten Meyer, Sol moved from $5 input and $30 output per million tokens in June to $2 and $10 for GPT-6.1 Sol in September. Luna moved from $1 and $6 to 10 cents and 50 cents. These are comparisons across model releases.

Did Anthropic lower its prices too?

Yes. The source reports that Opus moved from $5 input and $25 output per million tokens for Opus 5 to $4 and $20 for Opus 5.5. Anthropic also changed cache-read pricing for some models.

What should companies compare besides list prices?

They should review usage allowances, credit caps and expiry, seat minimums, overage rates and renewal terms. They should also test model performance on their own workloads because a lower rate does not by itself establish lower total costs.

Are OpenAI’s reported customer discounts available to every buyer?

The reports describe particular customer offers, not a standard discount for all buyers. The full terms and eligibility conditions for those deals are not public in the source material, so organizations would need to request and review their own written proposals.

Source: ThorstenMeyerAI.com