🔍 Read the full analysis: Falling AI Prices: What To Check In OpenAI And Anthropic’s Offers on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI and Anthropic have reduced prices on several AI models, while reporting describes additional discounts and changing customer choices. The reported 80% to 90% savings for DataCamp refers to expected overall costs, not a cut of that size from OpenAI’s list prices. Buyers should compare model performance, usage terms, credits and contract conditions alongside headline prices.
OpenAI and Anthropic have cut prices on several AI models since June, and OpenAI is also offering discounts to some enterprise customers, according to reporting cited by ThorstenMeyerAI.com on October 26. The changes could lower costs for companies using AI, but headline token prices do not show the full deal: usage allowances, seat minimums, credits and model performance can change the actual value.
Pricing pages reviewed by ThorstenMeyerAI.com showed OpenAI’s Sol model price moving 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. The figures compare successive models and pricing changes; they do not establish that an identical model became cheaper while remaining unchanged.
Anthropic’s Opus pricing shifted from $5 input and $25 output per million tokens for Opus 5 in July to $4 and $20 for Opus 5.5. The source also reports lower charges for cached input: OpenAI’s September change set GPT-6.1 Sol cache pricing at $0.10 per million tokens, while Anthropic reduced cache-read prices for Fable 5.1. Anthropic says those changes lower costs on typical and highly agentic workloads, but those workload estimates are the company’s own claims.
Enterprise deals described in reporting vary. The Information reported that OpenAI offered Replit a discount of more than 30% on a contract worth over $100 million a year, according to a person cited in the report. DataCamp CEO Jonathan Cornelissen said OpenAI pricing could reduce his company’s costs by 80% to 90%. That figure describes expected savings against DataCamp’s overall current setup, not a discount of that size from OpenAI’s public 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
What Lower AI Prices Change
Lower prices can make it less expensive for companies to run coding assistants, agents and other AI workloads, and can improve their position when renewing vendor contracts. The reported changes at Rippling illustrate that pricing may affect which systems employees use: 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 are figures attributed to Conrad, not a broader measure of market share.
For buyers, the practical comparison is the total cost of a workload, not just the price per million tokens. Cached-token rates can matter for systems that repeatedly process conversation history. Model quality also matters: a lower-priced successor may not perform equally well on a particular company’s tasks. The source says buyers should test their own workloads before treating a price reduction as an equivalent service at lower cost.
Contract terms can offset savings. Anthropic added capped, expiring API credits to its Max and Team plans, while its Enterprise plan has a 20-seat minimum. OpenAI, meanwhile, reduced the usage included in new $200 Pro subscriptions without changing the subscription price, according to the source. Customers who focus only on the monthly fee may miss changes to how much service that fee covers.
Products Worth Considering
Samsung 32-Inch Odyssey OLED G8 (G80SH) 4K 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 Accumulated
The changes took place over several months rather than in one announcement. OpenAI and Anthropic introduced new models and adjusted prices between June and September 2026. Data scientist Max Woolf published a timeline on October 25; ThorstenMeyerAI.com says it checked the listed figures against vendor announcements and pricing pages retrieved October 26.
Some reductions apply to new model versions, while others involve introductory prices becoming permanent or specific charges such as cache reads. That distinction matters when comparing prices over time: customers need to check which model, token type and date each figure refers to. A price comparison that combines different models or omits cache charges may not describe a customer’s actual bill.
Customer moves have also drawn attention. 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 that OpenAI was its preferred partner. A person familiar with the partnership told TechCrunch it was still in place, leaving the precise status less clear than the onstage remark suggested.
“I think ChatGPT is better now.”
— Amjad Masad, Replit CEO, speaking at TechCrunch Disrupt
Limits Behind the Savings Figures
The reported savings are not directly comparable. DataCamp’s 80% to 90% estimate concerns its expected total costs, while the Replit figure is a reported discount on a contract. The source does not provide the full contract terms, usage volumes or calculations behind either deal, so the effective per-workload savings cannot be independently assessed from the available details.
It is also unclear how widely enterprise discounts are available, how long they last or what commitments customers must make to receive them. The reported prices and plan terms may change after the October 26 review date. The source does not provide full details of the US bank deal, including the price or conditions attached to its expanded seat count.
Finally, lower prices do not establish that newer models match older ones on every task. Anthropic’s workload-cost estimates are vendor claims, and OpenAI’s descriptions of model capability are also vendor claims. Companies need to test quality, latency, reliability and total usage costs against their own work before drawing conclusions.
What Buyers Should Check Next
Companies considering renewals or new AI contracts can ask vendors to spell out discount duration, token rates, cache pricing, seat minimums, included usage and credit expiration. They should model costs using their own workload patterns, including how often systems reuse context, and compare those estimates with actual bills rather than relying on a headline savings percentage.
Buyers can also test more than one model on representative tasks and keep the ability to route work across providers where contracts and technical requirements allow. That approach can show whether a cheaper model meets quality requirements and reduce dependence on a single vendor. Individual subscribers should check whether usage limits or allowances have changed even when the monthly price has not.
The next indicators are updated vendor pricing pages, contract terms and customer reports that clarify whether discounts persist and how they affect real usage. Until those details are available, the public price cuts are clear, but the full scale and durability of enterprise savings remain unsettled.
Key Questions
Did OpenAI cut its prices by 90%?
Not according to the figures cited. The 80% to 90% figure was DataCamp’s CEO describing expected savings against the company’s total current setup. OpenAI’s listed token prices changed across models, but that is not the same as a 90% list-price cut.
How much did OpenAI’s Sol and Luna prices change?
For the models compared in the source, Sol moved 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. These are comparisons across model versions and dates.
Did Anthropic reduce all of its model prices?
The source describes several changes, including lower Opus 5.5 prices compared with Opus 5 and reduced cache-read charges for Fable 5.1. It does not establish that every Anthropic model or customer plan became cheaper.
What should a company compare beyond token prices?
Check included usage, cache charges, seat minimums, credit caps and expiration dates, as well as contract length and discount conditions. Test model quality and estimate costs with the company’s own workload before choosing a provider.
Source: ThorstenMeyerAI.com
