🔍 Read the full analysis: OpenAI And Anthropic Are Targeting Rival 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 have cut prices on several AI models in recent months, and reporting describes OpenAI offering discounts to customers using rival services. A reported 80% to 90% savings figure refers to one customer’s expected overall cost reduction, not a general discount off OpenAI’s list prices. Contract terms and usage allowances are changing alongside headline prices.
OpenAI and Anthropic have lowered prices for several AI models in recent months, as OpenAI also offers customer-specific discounts to win business from companies using rival services, according to reporting by The Information. The changes could reduce AI bills for some buyers, but reported savings vary by contract and workload, and lower list prices do not capture changing usage limits or other terms.
According to the vendors’ pricing pages and announcements reviewed by ThorstenMeyerAI.com on October 26, OpenAI’s GPT-6.1 Sol was listed at $2 per million input tokens and $10 per million output tokens, compared with $5 and $30 for the earlier Sol model in June. Its smaller Luna model went from $1 and $6 per million tokens to $0.10 and $0.50. These are prices for different model versions, not simply a reduction for an unchanged product.
Anthropic also cut prices across model releases. Its Opus line 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, while cache reads for the newer model were listed as 60% cheaper. Anthropic said Opus 5.5 costs 40% less on typical workloads; that is the company’s estimate, not an independently established result. Anthropic also said its Fable 5.1 model could cost less on certain workloads because of reduced cache-read pricing.
The customer offers described in coverage of an October 24 Information report include an OpenAI discount of more than 30% on a Replit contract reportedly worth over $100 million a year. DataCamp CEO Jonathan Cornelissen said the offer could reduce his company’s costs by 80% to 90%. That figure describes expected total savings against DataCamp’s existing setup, not an across-the-board discount from OpenAI’s published prices. A US bank was also reported to have received ChatGPT seats for less than $7 each; the source material does not provide further contract details.
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
Price Cuts Reshape AI Buying
For companies buying AI services, the cuts create room to renegotiate contracts and test competing models. A customer’s bill depends not just on the posted price per token, but also on how much it uses, which model performs its work, and what discounts or usage commitments apply. The reported DataCamp figure shows why an individual customer’s total savings should not be read as a general price reduction available to every buyer.
There are signs that customers are shifting workloads. 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%. Those figures are his account of one company’s usage, not a measure of broader market share. Still, the example suggests that price, model performance and contract terms can influence which provider businesses choose for particular tasks.
The competition also reaches individual users through changing plan allowances. The source reports that OpenAI reduced usage included with new $200 Pro subscriptions without changing the monthly price, while Anthropic introduced capped, expiring API credits for some plans. Buyers comparing services may need to review limits and credit rules as well as monthly fees and token prices.
Products Worth Considering
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How Model Prices Shifted
The reported price changes followed a series of model launches and revisions between June and September 2026. OpenAI discounted Luna in July, then announced lower prices for later Sol and Luna models in September. Anthropic introduced Sonnet 5 at an introductory price in June and made that price permanent in August, then released Opus 5.5 at a lower listed price than Opus 5. Max Woolf, a data scientist who tracks model prices, documented the sequence in an October 25 post. ThorstenMeyerAI.com said it checked the listed figures against vendor announcements and pricing pages retrieved October 26.
Some savings may come from cache pricing, not changes to standard input and output rates. Cached tokens can matter when an AI agent repeatedly processes the same conversation or working context. Woolf estimated that cached tokens account for 96% to 99% of token spending in agentic coding; this is his estimate, and actual costs depend on how a system is used. Anthropic’s claims that its newer models cost less on typical or highly agentic workloads are company estimates.
Reported customer relationships also require care in interpretation. Replit CEO Amjad Masad said at TechCrunch Disrupt on October 16 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, leaving the exact status unclear.
“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
Contract Terms Still Vary
The full terms of the reported OpenAI offers are not public in the source material. It does not give the details needed to compare each customer’s contract, such as duration, usage commitments, included services, or how savings were calculated. The reported 80% to 90% figure is one executive’s expectation, and should not be treated as a standard offer or a measured outcome across customers.
It is also unclear whether newer, cheaper models match the performance of earlier versions on every business task. The source reports that Anthropic’s Enterprise plan requires 20 seats, which may put it beyond the reach of some smaller firms, and that the company added capped, expiring API credits to Max and Team plans. It also says OpenAI changed usage included with new Pro subscriptions. The available material does not detail every plan’s limits or establish how widely these changes affect customers.
Replit’s relationship with Anthropic remains uncertain because accounts cited in the source differ. The company described OpenAI as its preferred partner while saying it remained model agnostic, and a person familiar with the partnership told TechCrunch it was still in place. No additional contract details are provided.
Buyers Compare Total Costs
Companies considering a switch will need to test models on their own workloads and compare total operating costs, rather than relying only on public token prices or headline discount claims. That means checking rate cards alongside usage caps, expiring credits, seat minimums, contract commitments and the cost of routing work between providers.
Further vendor announcements, revised pricing pages and customer disclosures may clarify how broadly the offers are available. The source material does not identify a specific next announcement or deadline. For now, the reported price changes and individual deals show a competitive push, but do not establish what terms other customers can obtain or how much they will save.
Key Questions
Are OpenAI and Anthropic cutting prices on existing models?
Some of the listed price reductions apply to newer model versions replacing earlier ones. The source reports other changes to introductory pricing and cache rates. A lower price for a successor model does not by itself show that the same model became cheaper or performs identically.
Does OpenAI offer every customer a 90% discount?
No such general discount is established by the source. The 80% to 90% figure came from DataCamp’s CEO describing expected overall savings against his company’s existing setup. A separate reported Replit offer was a discount of more than 30% on a contract said to be worth over $100 million annually.
Why might a customer’s actual AI bill differ from list prices?
Costs can depend on model choice, token usage, cached inputs, contract discounts and plan allowances. The source also reports changes to usage limits and credits, so customers need to review contract terms alongside per-token rates.
Are customers moving from Anthropic to OpenAI?
Some reported examples point to shifts, including Rippling’s CEO describing a change in the share of coding work handled by each provider. That is company-specific evidence, not proof of a market-wide trend. Replit’s relationship with Anthropic is also described inconsistently in the available reporting.
Source: ThorstenMeyerAI.com
