OpenAI And Anthropic’s Rivalry Is Reshaping AI Pricing

🔍 Read the full analysis: OpenAI And Anthropic’s Rivalry Is Reshaping AI Pricing on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

OpenAI and Anthropic have lowered prices for several AI models since June, and OpenAI is also offering discounts to win business customers, according to reporting cited by The Information. The changes are reshaping vendor choices and giving organizations more room to negotiate, though the widely repeated claim of 80% to 90% savings describes one customer’s expected total cost reduction—not a discount of that size from OpenAI’s list price.

Price comparisons compiled by data scientist Max Woolf and checked against vendor announcements and pricing pages show steep reductions across both companies’ model lines. OpenAI’s 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 fell from $1 and $6 to 10 cents and 50 cents per million tokens. Anthropic’s Opus line moved from $5 and $25 for Opus 5 in July to $4 and $20 for Opus 5.5; Anthropic also cut the price of cached reads for some models.

The changes involve new model versions as well as price adjustments. That distinction matters: a lower price for a successor does not by itself establish that it performs as well as the older model on a particular company’s tasks. Anthropic has said Opus 5.5 costs 40% less than Opus 5 on typical workloads, and that its Fable 5.1 model can cost about 25% less on typical workloads and up to about 45% less on highly agentic ones. Those are Anthropic’s estimates, not independent measurements.

The Information reported on October 24 that OpenAI was using discounts to compete for major customers. Among the examples, one person cited by the publication said OpenAI offered software company Replit a discount of more than 30% on a contract worth more than $100 million a year. DataCamp CEO Jonathan Cornelissen said the offer could reduce his company’s AI costs by 80% to 90% against its existing setup. That figure is an expected total saving, not a stated reduction from OpenAI’s published rates.

Lower Rates Change Vendor Choices

For businesses that spend heavily on AI, lower prices can reduce operating costs and make it easier to test more than one provider. The reported shift at software company Rippling illustrates how pricing can affect deployment: 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%. Conrad cited cost as a reason for the change. Those figures describe Rippling’s use, not a broader market-wide shift.

The headline price per token is only one part of the bill. Companies also need to account for cached-token rates, subscription usage limits, credits, minimum seat counts and contract-specific discounts. For individual subscribers, a monthly fee may stay the same even as the amount of usage included changes. Buyers who compare only list prices or subscription fees could miss a material change in what they receive.

Price competition can also influence which models developers build into products and workflows. But a lower quote does not establish that a model is a better fit or that total costs will fall by the same amount. The practical value depends on performance for the buyer’s tasks, actual usage and the terms of the agreement.

How the Price Changes Accumulated

The reductions came in a series of launches and pricing updates between June and September 2026. Anthropic introduced Sonnet 5 at an introductory price of $2 per million input tokens and $10 per million output tokens, then made that price permanent. OpenAI reduced the price of its Luna model in July and cut Sol’s price in August before September releases brought further lower rates. Anthropic also introduced Opus 5.5 and reduced cache-read pricing for some offerings.

Woolf published his comparison on October 25. The source article says its author checked the figures against vendor announcements and current pricing pages retrieved October 26. Model prices are not directly comparable without accounting for differences in model versions, task performance and how a customer uses them. In coding-agent workflows, repeated context can make cached-token costs particularly relevant, but the share of spending varies by workload.

There are also signs that commercial terms are changing alongside public rates. Anthropic added capped, expiring API credits to its Max and Team plans, according to the source material, while its Enterprise plan requires a 20-seat minimum. OpenAI, meanwhile, reduced the usage included with new $200 Pro subscriptions while keeping the monthly price unchanged. These terms affect what a customer can buy and use, even when the headline model rate receives the most attention.

“I think ChatGPT is better now.”

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

Contract Terms and Savings Vary

Many reported deal terms are not public. The Information’s article is paywalled, and the source account draws on its public summary and other coverage. The exact scope, duration and conditions of the reported customer discounts are not established here. A reported offer for a specific contract should not be treated as a standard rate available to every buyer.

Replit’s relationship with OpenAI is also described differently across accounts. Masad said the company was no longer working with Anthropic at the October 16 event, then 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. The precise status and scope of each relationship are not clear from the available accounts.

It is also unclear how much of the price decline reflects lasting list-price changes, temporary promotions, newer models or customer-specific contract terms. A company’s real savings depend on its workloads, usage patterns and service requirements. Vendors’ estimates of lower costs on typical or agentic workloads may not match results in a particular deployment.

Buyers Will Compare Total Costs

Businesses evaluating or renewing AI contracts can test competing models on their own workloads and compare the full cost of running them, including input and output tokens, cache use, usage caps, seats and expiring credits. They can also ask vendors to specify which prices are promotional, how long discounts last and whether usage terms can change during a subscription.

Further vendor announcements and customer disclosures may clarify whether the current reductions persist and how widely negotiated offers are available. For now, the confirmed public price changes show a sharp competitive push, while many commercial terms remain private. Buyers and subscribers will need to watch both the published rates and the amount of service included.

Key Questions

Did OpenAI cut its prices by 80% to 90%?

Not according to the reported figure. DataCamp CEO Jonathan Cornelissen said OpenAI’s offer could reduce his company’s total AI costs by 80% to 90% against its existing setup. That is not a stated discount of the same size from OpenAI’s list prices.

How much did OpenAI’s listed model prices fall?

In the comparisons provided, 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 per million tokens. These comparisons involve successive model versions.

Has Anthropic also lowered prices?

Yes. Opus moved from $5 and $25 per million input and output tokens for Opus 5 to $4 and $20 for Opus 5.5. Anthropic also made Sonnet 5’s introductory rates permanent and reduced cache-read prices for some models.

Why check terms beyond the token price?

Usage allowances, seat minimums and credits can change the total cost even when a listed rate or monthly subscription price does not. Buyers should compare actual expected usage and the contract’s limits, not only the headline price.

Are customers switching from Anthropic to OpenAI?

Some reported customer decisions point to shifts, including Rippling’s reported change in the share of coding work handled by each company’s models. But those are company-specific figures, not evidence of an industry-wide migration.

Source: ThorstenMeyerAI.com