A Look At The Prices And Terms In OpenAI And Anthropic’s AI Rivalry

🔍 Read the full analysis: A Look At The Prices And Terms In OpenAI And Anthropic’s AI Rivalry on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets 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 for several AI models since June, and OpenAI is also offering reported discounts to win business from some large customers. The October 24 reporting on those deals comes as subscription allowances, credit rules and seat requirements change, making the full cost of using each service harder to judge from list prices alone.

According to the vendors’ pricing pages and announcements, OpenAI’s larger Sol model 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 went from $1 and $6 to $0.10 and $0.50. Anthropic’s Opus pricing also fell: Opus 5 was listed at $5 per million input tokens and $25 per million output tokens in July, while Opus 5.5 was listed at $4 and $20. Anthropic also reduced the price of cached input for some models.

The price changes do not all represent the same model becoming cheaper. Several came with the release of successor models, so buyers need to test whether a newer model performs as well on their own work. The source’s review of vendor announcements and pricing pages, retrieved October 26, found that the listed figures matched the changes it tracked. Anthropic has said Opus 5.5 costs 40 percent less than Opus 5 on typical workloads; that is the company’s comparison, not an independently established result.

Separate customer deals may go beyond public pricing. The Information reported on October 24 that OpenAI was using discounts to pursue customers that use Anthropic. According to a person cited in that report, Replit received an offer of more than 30 percent off a contract worth over $100 million a year. DataCamp CEO Jonathan Cornelissen said the offer from OpenAI could reduce his company’s total AI costs by 80 to 90 percent. That figure describes expected total savings against the company’s current setup, not a 90 percent reduction in OpenAI’s published rates.

Why Buyers Are Rechecking AI Contracts

For companies with substantial model usage, lower prices can change the economics of building AI tools and coding assistants. The reported change at Rippling illustrates that vendors are competing not only for new business but also for workloads already assigned to a rival. CEO Parker Conrad said OpenAI’s share of the company’s coding work rose from 16 percent in May to 52 percent, while Anthropic’s fell from 57 percent to 18 percent. Those figures describe Rippling’s reported usage, not a market-wide shift.

The headline rate is only part of the bill. Buyers also need to compare input and output charges, cached-token pricing, included usage, contract discounts and the limits that apply when an allowance runs out. For individual subscribers, a monthly fee can stay unchanged even as the amount of usage included changes. For businesses, a discounted contract may offer substantial savings, but the comparison depends on the prior workload, current provider mix and contract terms.

That makes the current competition relevant to both procurement teams and users. Buyers may have more leverage to request revised terms, but they should test models on their own tasks and calculate costs over a representative period. A low published price does not by itself establish equivalent performance or a lower total bill.

A Fast-Moving Price Timeline

The price sequence began in June, when OpenAI previewed its GPT-5.6 Sol and Luna models at $5/$30 and $1/$6 per million input/output tokens. Anthropic introduced Sonnet 5 at an introductory $2/$10 rate later that month and made that rate permanent in August. OpenAI cut Luna’s rates in July and reduced Sol’s price in August, initially describing that Sol price as lasting three months.

In September, OpenAI priced GPT-6 Sol and Luna at lower rates, and GPT-6.1 Sol later reduced the cache price. Anthropic released Opus 5.5 at $4/$20 and introduced lower cache-read pricing for another model. The changes tracked by data scientist Max Woolf were reviewed against vendor announcements and pricing pages by the source article. These figures are dated snapshots, and vendor rates and plan terms can change.

Cached tokens are especially relevant to agentic coding, where a system may reuse earlier conversation context. Woolf estimates cached tokens account for 96 to 99 percent of token spending in agentic coding workloads. Anthropic has said its cache pricing can make certain workloads cheaper, including around 25 percent less on typical Fable 5.1 use and up to around 45 percent less on highly agentic workloads. Those estimates are Anthropic’s claims, and actual costs depend on how a customer uses the model.

“I think ChatGPT is better now.”

— Amjad Masad, Replit chief executive, speaking at TechCrunch Disrupt on October 16

What the Price Figures Leave Out

The reported customer offers do not establish that every buyer can obtain the same discounts. The Information’s article is paywalled, and the available accounts do not provide full contract terms or a complete account of how savings were calculated. The reported 80-to-90-percent estimate for DataCamp should not be treated as a standard offer available to other customers.

It is also unclear how long customer-specific discounts will last, what usage commitments they require, or how often companies will review them. Replit’s public statements about its relationship with OpenAI differ in emphasis: Masad called OpenAI its preferred partner, while a person familiar with the partnership told TechCrunch it remained in place. The source material does not resolve the precise scope of that arrangement.

Finally, price comparisons do not settle whether one model is a better fit. Performance can vary by task, and changes to included usage or credit expiration can affect total cost even when a monthly subscription price stays the same. Buyers will need their own usage data to make a reliable comparison.

What Buyers Should Check Next

Companies considering a renewal or new AI contract can ask vendors to quote against their actual workload and spell out usage caps, overage rates, credit expiration, seat minimums and any minimum commitments. They can also compare more than one provider and test models on representative tasks rather than relying solely on vendor benchmarks or introductory prices.

Users on paid plans should review what their subscription includes, especially if they signed up before a change to usage allowances. The source reports that OpenAI reduced the usage included in new $200 Pro subscriptions without changing the price. Anthropic added capped, expiring API credits to its Max and Team plans, while its Enterprise plan requires 20 seats. These conditions can matter as much as the monthly fee for smaller teams.

The next useful indicators will be updated vendor pricing pages, plan terms and further reporting on customer contracts. Until those details are public, the reported discounts show that some large buyers are negotiating lower costs, but not what a typical customer can expect to pay.

Key Questions

Did OpenAI offer DataCamp a 90 percent discount?

Not according to the source material. DataCamp CEO Jonathan Cornelissen described an expected 80-to-90-percent reduction in total costs against the company’s current setup. That is not the same as a 90 percent cut to OpenAI’s list price.

How much did OpenAI’s listed model prices fall?

From June to September, the listed price for the larger Sol line moved from $5/$30 to $2/$10 per million input/output tokens. Luna moved from $1/$6 to $0.10/$0.50. These are prices for different model versions, not necessarily the same models at reduced rates.

Are companies already moving work between the providers?

Rippling CEO Parker Conrad said OpenAI’s share of the company’s coding work rose from 16 percent in May to 52 percent, while Anthropic’s share fell from 57 percent to 18 percent. Those numbers are specific to Rippling.

What should buyers compare besides the listed token price?

They should check input and output rates, cached-token charges, included usage, caps, overage fees, credit expiration, seat minimums and contract commitments. They should also test model performance on their own tasks.

Source: ThorstenMeyerAI.com