🔍 Read the full analysis: OpenAI And Anthropic Are Chasing AI Customers With Lower Prices on ThorstenMeyerAI.com
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
TL;DR
OpenAI and Anthropic cut several model prices between June and September 2026, and reporting describes new discounts aimed at winning large customers. A claim that one buyer expects savings of 80% to 90% refers to its projected overall costs, not a discount of that size off OpenAI’s list prices. Buyers also need to track seat minimums, usage allowances and credit terms.
OpenAI and Anthropic have reduced prices for several AI models since June, and OpenAI is offering customer-specific discounts as it competes for business reportedly held by Anthropic. The development gives companies more room to negotiate, but reported savings vary by customer and should not be confused with the vendors’ public list prices.
According to the vendors’ pricing pages, OpenAI’s Sol model line 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 line fell from $1 and $6 to $0.10 and $0.50 per million input and output tokens. These figures compare successive models and do not mean every customer’s bill fell by the same amount.
Anthropic also lowered prices across model releases. Opus moved from $5 per million input tokens and $25 per million output tokens for Opus 5 in July to $4 and $20 for Opus 5.5. Anthropic also reduced the price of cached input tokens on some models. Such cache pricing can affect repeated requests in coding and other agent workflows, though the actual savings depend on how a customer uses the model.
Customer-specific offers are separate from those public prices. The Information reported on October 24 that OpenAI is using discounts to pursue customers associated with Anthropic. The source material describes an offer of more than 30% off for Replit on a contract reportedly worth more than $100 million annually. It also reports that a U.S. bank received 500 ChatGPT seats at no additional charge. Contract details and terms for these offers have not been made public.
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, More Contract Leverage
The cuts matter to organizations whose AI bills rise with model usage. A lower per-token rate can reduce costs directly, while vendor competition may give large buyers leverage to negotiate discounts, additional seats or other contract terms. The reported examples suggest that OpenAI is competing not only on public pricing but also through deals tailored to individual customers.
Still, the headline price is only one part of the bill. Companies need to compare input and output rates, cached-token charges, usage caps, seat minimums and credit expiry rules. A lower-priced model may also perform differently on a company’s work. The source material notes that many price comparisons involve successor models replacing earlier ones, so buyers should test quality and cost on representative tasks rather than assume equivalent performance.
For individual subscribers, the relevant comparison is also changing: a monthly fee can stay the same even as included usage changes. For businesses, the reported offers point to a practical opportunity to review existing agreements, but savings should be measured against current spending and actual contract terms, not an eye-catching percentage alone.
Products Worth Considering
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
As an affiliate, we earn on qualifying purchases.
How the Price Cuts Unfolded
The reductions came through a sequence of model launches and pricing changes during 2026. OpenAI lowered Luna pricing in July and cut Sol pricing in August before introducing new GPT-6 models at lower rates in September. Anthropic set introductory Sonnet pricing as permanent in August and released lower-priced Opus 5.5 in September. A pricing review in the source material says its figures matched vendor announcements and pricing pages checked on October 26.
Some reductions concern cached input rather than the standard input and output rates. Cached tokens are reused portions of an earlier prompt, and vendors charge differently for them. Anthropic has attributed lower typical-workload costs to such changes, including estimates for its own models. Those workload savings are vendor estimates, not guarantees for every customer.
The commercial competition is reflected in reported shifts at customer companies. Rippling CEO Parker Conrad said OpenAI’s share of his company’s coding work rose from 16% in May to 52%, while Anthropic’s fell from 57% to 18%. He cited cost as a reason. Those figures describe Rippling’s reported coding use, not overall market share.
“I think ChatGPT is better now.”
— Amjad Masad, Replit CEO, speaking at TechCrunch Disrupt on October 16
What the Deals Do Not Reveal
The full terms of the reported customer offers are not public, including how long discounts last, which models or usage levels they cover, and whether they come with commitments or renewal conditions. The reported 80% to 90% DataCamp savings are an executive’s expectation relative to the company’s current setup; the available information does not provide enough detail to independently calculate that estimate.
Replit’s relationship with OpenAI also remains described differently in public comments and reporting. Masad called OpenAI its preferred partner while saying Replit remained model agnostic, and a person familiar with the partnership told TechCrunch it was still in place. The precise scope of the arrangement is unclear. More broadly, list prices cannot establish how much a typical customer saves, because usage patterns, negotiated rates and plan limits vary.
What Buyers Should Track
Companies considering a switch or renewal can compare vendors using their own workloads, including model quality, token use, cached-input rates and support requirements. They can also ask vendors to spell out discount duration, seat counts, usage allowances and credit expiration in writing, then compare the full expected bill rather than a single quoted rate.
Customers should watch for further pricing and plan changes from both vendors, as well as clearer public details about the reported contracts. The next useful evidence will be whether discounts persist at renewal and whether lower per-token rates translate into lower total spending for customers with real production workloads.
Key Questions
Did OpenAI cut its prices by 90%?
No, not according to the information available. The 80% to 90% figure refers to DataCamp CEO Jonathan Cornelissen’s expected reduction in the company’s overall costs under an offer. It is not a 90% cut to OpenAI’s public list prices.
How much did OpenAI’s listed model prices fall?
For the model lines cited, OpenAI’s Sol rates went from $5/$30 to $2/$10 per million input/output tokens between June and September. Luna went from $1/$6 to $0.10/$0.50. The figures compare successive models, and a customer’s actual savings depend on usage and contract terms.
Did Anthropic also lower prices?
Yes. Anthropic’s cited Opus pricing moved from $5/$25 to $4/$20 per million input/output tokens between Opus 5 and Opus 5.5. The company also changed cache pricing on some models.
Are companies already switching from Anthropic to OpenAI?
There are reports of shifts at particular customers, not evidence that all customers are moving. Rippling’s CEO reported a change in the share of coding work handled by each vendor, while Replit described OpenAI as a preferred partner and said it remained model agnostic.
What should buyers compare besides list prices?
Compare the total expected cost, including input and output tokens, cached-token rates, usage limits, seat minimums, credits and renewal terms. Test model performance on the work the organization actually needs to do.
Source: ThorstenMeyerAI.com
