🔍 Read the full analysis: OpenAI And Anthropic Cut Prices As They Compete For AI Users on ThorstenMeyerAI.com
Get smart everyday buys 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 prices across several models between June and September 2026, and October reporting describes OpenAI offering discounts to win business from Anthropic. The reported 80% to 90% savings refer to one customer’s expected total costs, not a general list-price reduction; contract terms and usage limits also affect what buyers pay.
OpenAI and Anthropic cut prices for several AI models between June and September, while an October report described OpenAI offering discounted deals to attract customers from Anthropic, a development explored in the CFO’s new operating system. The changes matter to companies choosing AI providers, but the headline figures do not represent a single across-the-board discount: prices, model capabilities, contract terms and included usage all vary.
According to the vendors’ pricing information and announcements, OpenAI’s GPT-6.1 Sol costs $2 per million input tokens and $10 per million output tokens, down from $5 and $30 for the earlier Sol model in June. Its Luna model’s listed price fell from $1 and $6 to $0.10 and $0.50. These comparisons are between different model releases, not simply a reduction for an unchanged model.
Anthropic also lowered prices across releases. Its Opus line 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 also cut cache-read pricing for some models. The source report attributes claims that these changes lower costs on typical or agent-heavy workloads to Anthropic; actual savings depend on how customers use the models.
Separate from public list prices, The Information reported on October 24 that OpenAI was offering customer-specific discounts, amid a broader shift toward consulting-style deals. Its reported examples include a discount of more than 30% on a Replit contract valued at more than $100 million per year, and seats priced below $7 for a U.S. bank. DataCamp CEO Jonathan Cornelissen described expected total cost savings of 80% to 90% from OpenAI’s offer, according to the report. That figure is an estimate against DataCamp’s existing setup, not a 90% cut to OpenAI’s published prices.
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
For companies that spend heavily on AI, falling token prices and competition for large accounts may create room to renegotiate contracts or shift workloads between providers. The reported deals suggest that list prices are only one part of enterprise purchasing: negotiated rates, seat counts and included usage can change the effective cost substantially.
The changes also make comparisons harder. A newer model may cost less but perform differently on a company’s particular tasks. In addition, token costs are not the full bill when subscriptions have caps, credits expire, or a contract requires a minimum number of seats. Buyers need to compare the same workloads and terms rather than relying on a single advertised price.
For individual users, lower-cost models may make AI more accessible, but a subscription’s price can remain unchanged even as its usage allowance changes. Customers should check what a plan includes, not just its monthly fee. The reported competition could benefit buyers, but the scale and duration of those benefits depend on their contract and actual usage.
Products Worth Considering
Dell UltraSharp 32 U3226QER 4K Thunderbolt Hub Monitor
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
A Summer of Model Price Changes
The reductions followed a series of launches and price adjustments during 2026. The source material, citing vendor announcements and pricing pages checked on October 26, lists OpenAI model prices changing in June, July and September. Anthropic launched Sonnet 5 at an introductory price in June and made that price permanent in August, then released Opus 5.5 in September at lower listed rates than Opus 5.
These comparisons require care because many price changes came with new model versions, rather than a vendor charging less for precisely the same system. OpenAI described GPT-6.1 Sol as offering “near-Astra intelligence for a fifth of the price,” according to the source. Whether a successor provides comparable results for a particular task must be tested by the customer; the pricing figures alone do not establish equivalent performance.
Some savings involve cached tokens, which can matter when software agents repeatedly process earlier conversation or task context. Data scientist Max Woolf estimated that cached tokens account for 96% to 99% of token spend in agentic coding. That estimate concerns agentic coding workloads and should not be treated as a share that applies to every customer or use of AI.
Deal Terms and Savings Vary
The reported discounts are not a public price schedule, and the available source material does not provide full contract terms for each customer. It is unclear how broadly OpenAI is offering the deals, how long they last, or whether they require usage commitments or other conditions. The Information article is paywalled; the cited details come from its public summary and subsequent accounts.
Some customer details also remain qualified. Replit’s CEO described OpenAI as a preferred partner and said the company remained model agnostic, while a person familiar with the partnership told TechCrunch that it was still in place. The precise scope of that relationship is not established here. The reported 80% to 90% savings for DataCamp are an executive’s expectation, not a verified result across customers.
Subscription terms are another moving part. The source says Anthropic added capped, expiring API credits to Max and Team plans and that its Enterprise plan requires 20 seats. It also says OpenAI reduced usage included with new $200 Pro subscriptions without changing the price. The full scope, timing and effect of those plan changes are not detailed in the supplied reporting.
Watch Renewals and Usage Limits
Companies weighing a change should compare their own workload costs, model performance and contract terms, including any seat minimums, usage caps, credit expiry dates and cache pricing. Renewals and new enterprise deals will show whether reported discounts are isolated negotiations or part of a wider pricing strategy.
Customers can also track whether the vendors make further changes to public prices or subscription allowances. For now, the confirmed pattern is a run of lower model prices and reported negotiated offers; the scale and staying power of the savings remain uncertain. Any comparison should distinguish published rates from customer-specific discounts and evaluate models on the work they are expected to perform.
Key Questions
Did OpenAI cut its prices by 90%?
No general 90% list-price reduction is established here. The 80% to 90% figure was DataCamp CEO Jonathan Cornelissen’s estimate of expected savings against his company’s existing setup, as reported by The Information.
How much did OpenAI’s listed model prices fall?
The source says GPT-6.1 Sol was listed at $2 per million input tokens and $10 per million output tokens, compared with $5 and $30 for an earlier Sol model in June. The comparison is between model releases, not the same model at two prices.
Did Anthropic lower its model prices too?
Yes. The source reports that Opus pricing 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 also changed cache-read pricing for some models.
Are businesses switching from Anthropic to OpenAI?
Some reported customer usage and negotiations point in that direction, but the evidence does not establish a general shift. Rippling’s CEO reported a change in the models used for its coding work, and reporting described OpenAI discounts for customers including Replit. Replit’s CEO said the company remained model agnostic.
What should AI buyers compare besides list price?
They should compare costs for their own workloads, model performance, negotiated rates, included usage, seat minimums, credit expiry and any usage caps. A lower published token price does not by itself establish a lower total bill.
Source: ThorstenMeyerAI.com
