What’s Behind The OpenAI And Anthropic AI Price War?

🔍 Read the full analysis: What’s Behind The OpenAI And Anthropic AI Price War? on ThorstenMeyerAI.com

Age 18–24?Offer from Amazon

Prime made for students and young adults

• Fast, free delivery for dorm and study essentials
• Prime Video and Amazon Music included
• Member-only deals

Try Prime for Young Adults →Free trial for eligible 18–24 year olds

As an affiliate, we earn on qualifying purchases.Hide

OpenAI and Anthropic have lowered prices on several AI models since June, as OpenAI also offers reported discounts to some large customers and the companies adjust subscription and API terms. The changes are giving buyers more room to negotiate, but an 80% to 90% savings claim cited in recent coverage refers to one customer’s expected total cost reduction, not a broadly available discount off OpenAI’s list prices.

According to the vendors’ pricing pages, 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 went from $1 and $6 to $0.10 and $0.50. These prices compare different model versions, rather than showing the same model becoming cheaper.

Anthropic also reduced prices across model releases. Its Opus line moved from $5 input and $25 output per million tokens for Opus 5 in July to $4 and $20 for Opus 5.5. Anthropic also reduced cache-read costs on some models. The company says its newer models can cost less on typical workloads, but those performance and savings estimates are vendor claims, and the result for a given customer depends on how the model is used.

The Information reported on October 24 that OpenAI was offering discounts to attract customers from Anthropic. The reported examples include a discount of more than 30% for Replit on a contract worth more than $100 million a year, according to a person cited in the report, and lower-priced ChatGPT seats for a U.S. bank. DataCamp CEO Jonathan Cornelissen said OpenAI’s offer could reduce his company’s costs by 80% to 90%; that figure describes his expected savings against its existing setup, not a list-price reduction available to all buyers.

Lower Rates, More Contract Leverage

For companies buying large amounts of AI, falling rates and vendor competition can improve negotiating leverage. But per-token list prices are only one part of the bill. Discounts may depend on contract size, volume, model choice or other terms, and customers need to compare the cost of completing their actual work rather than assuming that a lower rate will produce equivalent results.

Subscription changes can also offset headline price cuts. The source reports that OpenAI reduced the usage included in new $200 Pro subscriptions without changing the monthly fee. Anthropic added capped, expiring API credits to its Max and Team plans, while its Enterprise plan requires at least 20 seats. These terms can affect smaller teams and individual users even when the displayed subscription price stays the same.

For buyers, the competitive shift also makes vendor dependence a practical consideration. Companies that can test models across providers may have more flexibility when rates or service terms change. However, switching is not automatic: teams need to check model performance, reliability, integration costs and contract restrictions alongside the price.

How the Price Cuts Accumulated

Data scientist Max Woolf published a timeline of OpenAI and Anthropic price changes on October 25. The source article says it checked the figures against vendor announcements and pricing pages retrieved on October 26. The sequence includes OpenAI’s June prices, subsequent cuts to Luna and Sol, Anthropic’s introductory Sonnet pricing becoming permanent, and September releases with lower listed rates or cache costs.

One less visible part of the competition is cached input pricing. Coding agents can repeatedly process prior conversation context, so cache rates may affect the cost of some workloads. Woolf estimates that cached tokens make up 96% to 99% of token spending in agentic coding; that is his estimate, not a universal measure. Anthropic has said cache changes reduce costs on typical and highly agentic workloads, but customers should test those claims against their own usage patterns.

Customer use appears to be shifting in at least one cited example. 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%. Conrad attributed the shift to cost. Those figures describe Rippling’s reported experience, not a broader market-wide measure.

What the Savings Figures Cover

The reported customer discounts are not a public price schedule. The source does not provide the full contracts, eligibility rules or service conditions for the reported Replit and bank deals. It is also unclear how broadly OpenAI is offering comparable terms, or how long any customer-specific discounts will last.

The 80% to 90% figure is especially easy to misread. It is DataCamp’s CEO’s estimate of expected total savings against the company’s current setup, as reported in the source material. It is not evidence that OpenAI cut its public prices by that amount. The underlying costs, workloads and assumptions behind the estimate are not fully detailed here.

Some apparent reductions compare newer models with older ones, so price alone does not establish equal capability or value. The source also reports conflicting indications about Replit’s relationship with Anthropic: Masad said it was no longer working with Anthropic at one event, while a person familiar with the partnership told TechCrunch it remained in place. The company’s actual provider mix and contract status are not fully clear.

Track Terms Alongside Model Rates

Companies evaluating new deals can compare current vendor price pages with their invoices and contract terms, then test competing models on representative tasks. A useful comparison should account for input and output tokens, cached-token rates, usage limits, seat minimums, credits and any discount conditions, as well as the quality of the results. Buyers with renewal dates approaching may have an opportunity to request updated terms, but no particular discount is guaranteed.

Users on paid plans should check whether usage allowances or credit rules have changed, not just whether the monthly fee has moved. OpenAI and Anthropic may continue revising rates as models and offers change. The next clear evidence of how the competition is affecting buyers will come from updated pricing pages, disclosed contract terms and customer reports that distinguish list prices from negotiated savings.

Key Questions

Are OpenAI and Anthropic cutting prices on the same models?

No. The reported comparisons include newer model versions replacing older ones, as well as changes to cache pricing and introductory rates. A lower rate does not by itself show that two versions perform equally on a buyer’s tasks.

Does OpenAI offer everyone a 90% discount?

No such general discount is established by the source material. The 80% to 90% figure was DataCamp CEO Jonathan Cornelissen’s estimate of expected total savings against his company’s existing setup, not a public discount off OpenAI’s list prices.

Why do cache prices matter?

Some workloads repeatedly reuse earlier context, which can make cached input rates relevant to their total cost. The effect varies by workload, so customers should measure their own usage rather than apply a single estimate universally.

What should businesses compare before signing an AI contract?

Compare the likely total cost for real workloads, including model rates, cache pricing, included usage, credits, seat minimums, discount conditions and contract duration. Also test output quality and account for implementation or switching costs.

Source: ThorstenMeyerAI.com