🔍 Read the full analysis: OpenAI And Anthropic’s Price Fight: The Details AI Customers Should Check on ThorstenMeyerAI.com
Get hardware and tech essentials 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 lowered several model prices in 2026, and reporting describes OpenAI offering large customer-specific discounts to win business from Anthropic. The widely cited 80% to 90% savings figure describes one customer’s expected overall cost reduction, not a general discount off OpenAI’s list prices. Buyers should compare actual usage, contract terms and model performance, not just headline rates.
OpenAI and Anthropic have cut prices for several AI models since June, and a report published October 24 says OpenAI is offering targeted discounts to win customers from Anthropic. The developments could lower bills for businesses, but the widely repeated claim of 80% to 90% savings refers to one executive’s expected overall cost reduction—not a general discount on OpenAI’s published prices.
Pricing figures compiled from the vendors’ pages and announcements show OpenAI’s Sol models moving 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 Luna line moved from $1 and $6 to $0.10 and $0.50. These are prices for different model releases, rather than a simple reduction to an unchanged model.
Anthropic also reduced rates. Its Opus line went 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 has also described lower costs from changes to prompt caching. The company says Fable 5.1 can cost about 25% less on typical workloads and up to about 45% less on highly agentic workloads; those estimates are Anthropic’s, and actual savings depend on how a customer uses the models.
The Information reported on October 24 that OpenAI is using discounts to compete for Anthropic customers. The reported examples include a discount of more than 30% on a Replit contract valued at over $100 million a year, and a US bank receiving seats for under $7 each, according to accounts of the report. DataCamp CEO Jonathan Cornelissen said OpenAI’s offer could reduce his company’s costs by 80% to 90%. That figure describes expected total savings against DataCamp’s current setup, not a discount available to all buyers.
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
Contract Costs Beyond Token Rates
Falling list prices give customers a reason to revisit contracts, but they do not by themselves show what a business will pay. Token volume, cached-token pricing, seat counts, included usage, and overage rules can change the final bill. A low per-token rate may matter less if a plan includes fewer requests or imposes minimum seats.
Reported customer movement suggests price and performance are already influencing vendor choices. Rippling CEO Parker Conrad said OpenAI’s share of his company’s coding work rose from 16% in May to 52%, while Anthropic’s share fell from 57% to 18%. Those figures describe one company’s use, not a broad market share measure. For buyers, the practical question is whether a model performs well on their own tasks at a lower total cost.
Terms can also shift without a change in the advertised monthly fee. The source reports that OpenAI reduced included usage for new $200 Pro subscriptions, while Anthropic added capped, expiring API credits to some plans and has a 20-seat minimum for Enterprise. The details make it harder to compare plans using price labels alone. Businesses and individual subscribers alike need to check what usage is included and when limits apply.
Products Worth Considering
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K Gaming Monitor
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
A Four-Month Price Sequence
The cuts followed a rapid series of model launches and price changes between June and September 2026. OpenAI reduced Luna pricing in July and later announced lower rates for newer Sol and Luna models. Anthropic set introductory pricing for Sonnet 5 as permanent, then released Opus 5.5 at a lower listed rate than Opus 5. The comparison is between successive models, so it does not establish that an older model’s price was cut while its capabilities stayed constant.
Data scientist Max Woolf published a timeline of the reductions on October 25. The source article says it checked those figures against vendor announcements and pricing pages retrieved October 26. It also highlights prompt-cache prices, which can affect AI agents that repeatedly send prior conversation context. Woolf estimates cached tokens account for 96% to 99% of token spending in agentic coding. That is an estimate, not a universal rate for all customers or workloads.
In the reported customer examples, the distinction between price list and negotiated deal is central. Replit CEO Amjad Masad said onstage 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. TechCrunch also reported that a person familiar with the partnership said it remained in place.
“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 The Information
Savings and Terms Still Vary
The exact terms of the customer-specific offers are not public in the source material, including how long discounts last, what usage they cover, and whether they depend on volume or contract length. The Information article is paywalled; the source account says its summary was checked against reporting by The Decoder and all-ai.de. The reported figures should be treated as examples, not a standard offer available to every customer.
It is also unclear how much of each reported saving comes from lower rates, model selection, negotiated terms, or changes in how work is routed. The source does not provide full contract documents or comparable workload tests. Vendor claims about savings from caching and successor-model costs likewise may not match results for a particular customer. OpenAI and Anthropic’s published rates are only one part of the calculation.
Some plan details may also change over time. The source reports changes to included usage and credit rules, but does not provide complete terms for every plan or say whether all existing subscribers are affected. Customers should verify current contract and plan conditions directly before comparing offers.
Check the Next Contract Renewal
Businesses reviewing AI spending can use the price changes as a prompt to request updated quotes, but should compare the same workload across providers. That means checking input and output rates, cache pricing, included credits, expiration dates, seat minimums, usage caps, and charges beyond the allowance. Testing model quality on representative tasks matters too: a cheaper successor may not produce equivalent results for every use.
Customers should also ask whether a quoted discount applies throughout the contract term and what happens when promotional pricing ends. For individual users, the monthly fee may stay the same while usage allowances change, so the current plan limits deserve a separate check. The immediate next milestone is the next vendor pricing or plan update, alongside any further reporting that clarifies the terms of enterprise deals.
Key Questions
Did OpenAI offer everyone a 90% discount?
No. The 80% to 90% figure refers to DataCamp’s expected total cost savings against its existing setup, as attributed to CEO Jonathan Cornelissen. It is not a general discount on OpenAI’s published prices.
How much did OpenAI’s listed model prices fall?
The source comparison says OpenAI’s Sol rates 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. Luna moved from $1 and $6 to $0.10 and $0.50. These figures compare different model releases.
Did Anthropic also lower prices?
Yes. The source says Opus moved from $5 per million input tokens and $25 per million output tokens for Opus 5 to $4 and $20 for Opus 5.5. Anthropic has also said caching changes can reduce costs on some workloads; those savings depend on usage.
What should AI customers compare besides list prices?
Compare the actual workload cost, model performance, cached-token rates, usage allowances, seat minimums, credit limits and expiry dates, and any charges beyond included usage. Ask whether negotiated rates last for the full contract term.
Source: ThorstenMeyerAI.com
