🔍 Read the full analysis: OpenAI And Anthropic’s AI Price War: What Subscribers Need To Know on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI and Anthropic have cut prices on several models since June, and reporting points to new discounts aimed at large customers. The often-cited 80% to 90% savings figure describes one customer’s expected total cost reduction, not a general OpenAI list-price discount; plan limits and contract terms also matter.
OpenAI and Anthropic have cut prices on several AI models since June, as OpenAI also offers reported discounts to win business from major customers. The changes could lower costs for subscribers and companies, but the widely cited 80% to 90% savings figure is not a general list-price cut, and monthly fees do not tell the whole story when usage limits and contract terms can change.
According to the vendors’ pricing pages, OpenAI’s Sol model moved from $5 input and $30 output per million tokens in June to $2 and $10 for GPT-6.1 Sol in September. Its Luna model moved from $1 and $6 to $0.10 and $0.50 per million tokens. These are prices for different model releases; they do not show that an unchanged model became cheaper.
Anthropic also reduced prices across releases. Opus moved from $5 input and $25 output per million tokens for Opus 5 to $4 and $20 for Opus 5.5. The source account also reports lower prices for cached input on some offerings. Anthropic said Fable 5.1’s cache changes reduce costs by around 25% on typical workloads and up to around 45% on highly agentic ones. Those estimates are the company’s claims, not independent measurements.
Customer-specific deals are separate from published prices. The Information reported that OpenAI offered Replit a discount of more than 30% on a contract worth over $100 million annually. DataCamp CEO Jonathan Cornelissen said he expected OpenAI pricing to reduce his company’s costs by 80% to 90%. That figure refers to his expected total savings against the company’s existing setup, not a discount of that size off OpenAI’s standard price list.
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
How Lower Prices Change AI Bills
Lower per-token rates can reduce costs for businesses that run coding assistants, agents or other high-volume AI workloads. But a headline price alone is a weak guide: model choice, cached-token pricing, actual usage and negotiated terms all affect the final bill. A newer model may also perform differently on a company’s tasks, so the cost comparison is useful only alongside testing.
For individual subscribers, the monthly charge may stay the same even as the included allowance changes. The source account says OpenAI reduced usage included in new $200 Pro subscriptions without changing the subscription price, while Anthropic added capped, expiring API credits to Max and Team plans. That makes it important to check what the plan includes, not just what it costs.
For buyers negotiating enterprise agreements, the reported deals suggest that vendors may offer terms unavailable on public pricing pages. Companies may have room to request revised rates, but savings reported by one buyer should not be treated as a standard offer or a promise to other customers.
Products Worth Considering
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
LG UltraGear 27-Inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDMI 2.1
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
As an affiliate, we earn on qualifying purchases.
A Summer of Model Price Cuts
Data scientist Max Woolf published a timeline of model price changes on October 25. The account supplied for this article says its author checked the figures against vendor announcements and pricing pages retrieved October 26. The timeline records several moves from June through September, including OpenAI’s reductions to Luna and Sol and Anthropic’s introductory Sonnet 5 price becoming permanent.
Many listed reductions came with new model releases rather than cuts to the same model. That distinction matters when comparing value: a lower price does not by itself establish equivalent performance. The account also notes that cached input can be a large share of spending in agentic coding, but the estimate that cached tokens make up 96% to 99% of token spend is attributed to Woolf, not presented as a universal result for every customer.
The Information reported on October 24 that OpenAI was using discounts to attract customers associated with Anthropic. The reported examples include a changing distribution of coding work at Rippling and the Replit offer. These are specific company accounts and reported deal terms, not evidence that all customers are changing vendors or receiving similar discounts.
“I think ChatGPT is better now.”
— Replit CEO Amjad Masad, speaking at TechCrunch Disrupt
Deal Terms and Limits Still Vary
Public information does not establish the full terms of the reported customer contracts, including how long discounts last, which workloads qualify or whether the rates apply across a customer’s usage. The reported 80% to 90% savings is an individual executive’s expectation, and the source does not provide a comparable, independently verified cost audit.
It is also unclear how broadly the reported offers are available. Details of the US bank deal mentioned in the source are incomplete, and the account does not give enough information to assess its pricing or contract conditions. Customer migration claims are similarly specific: Rippling’s CEO described a shift in coding-work share, while Replit’s public comments and reported partnership details do not establish that every part of its work has moved to one provider.
Finally, lower rates do not guarantee lower total bills. Usage caps, expiring credits, seat minimums, caching behavior and the performance of successor models can change the result. Subscribers should check the terms that apply to their own plan or contract.
What Buyers Should Check Now
Companies reviewing AI spending can compare current invoices with vendor quotes and test the relevant models on their own workloads before changing providers. They should ask vendors to spell out token rates, cache pricing, usage allowances, seat minimums, credit expiry and renewal terms, and to clarify whether a quoted discount applies to all usage or only a defined contract.
Individual users should review plan pages and account notices for changes to included usage, rather than assuming the same monthly price means the same allowance. The next useful evidence will be updated vendor pricing and clearer disclosures of contract terms and customer outcomes. No further pricing move or company-wide change is confirmed in the supplied material.
Key Questions
Did OpenAI cut prices by 90%?
No general list-price cut of 90% is established. DataCamp’s CEO described expected total savings of 80% to 90% against his company’s existing setup. Separately, The Information reported an OpenAI offer of more than 30% off a Replit contract.
How much did OpenAI’s Sol and Luna prices change?
According to the pricing figures cited in the source, Sol went 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 $0.10 and $0.50. The comparisons involve different model releases.
Did Anthropic lower Opus prices?
The cited prices for Opus changed from $5 input and $25 output per million tokens for Opus 5 to $4 and $20 for Opus 5.5. Anthropic also reported lower costs from cache changes on some workloads, but those savings estimates are the company’s own figures.
What should subscribers check before renewing?
Check the current usage allowance, seat requirements, credit caps and expiry dates, as well as the monthly price. Business customers should also confirm which rates apply to their workload and how long negotiated discounts last.
Source: ThorstenMeyerAI.com
