The AI Price War: OpenAI And Anthropic Slash Prices And Fight For Each Other’s Customers. Read The Fine Print Before You Sign
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: The AI Price War: OpenAI And Anthropic Slash Prices And Fight For Each Other’s Customers. Read The Fine Print Before You Sign on ThorstenMeyerAI.com

Before you orderOffer from Amazon

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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI and Anthropic have cut prices for several AI models since June, and reporting describes OpenAI offering negotiated discounts to some business customers. The headline claim of 80% to 90% savings refers to one company’s expected total costs, not a general list-price discount. Customer usage, subscription limits and contract terms are also changing.

OpenAI and Anthropic have reduced prices for several AI models since June, while OpenAI is also offering some businesses negotiated discounts to win or expand contracts, according to reporting by The Information cited in the source material. The changes are reshaping costs for companies buying AI, but reported savings vary by customer and do not amount to a universal 80% to 90% cut in OpenAI’s published prices.

According to vendor pricing pages checked on October 26 and a price history compiled by data scientist Max Woolf, OpenAI’s Sol model fell 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 smaller Luna model moved from $1 and $6 to $0.10 and $0.50 per million input and output tokens. These figures compare different model releases, rather than a single model being discounted throughout.

Anthropic also cut prices across several launches. 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 lowered cache-read prices for some offerings. The company says those changes reduce costs on typical workloads, with larger savings on highly agentic tasks; those workload estimates are vendor claims, not independent measurements.

The Information reported on October 24 that OpenAI is using discounts to compete for customers. One reported offer to Replit exceeded 30% on a contract valued at more than $100 million a year, according to a person cited by the publication. DataCamp CEO Jonathan Cornelissen said OpenAI’s offer could reduce his company’s costs by 80% to 90%. That figure describes expected savings against DataCamp’s existing setup, not a discount of that size from OpenAI’s standard rates.

At a glance
reportWhen: Price changes reported from June throug…
The developmentA series of model price cuts by OpenAI and Anthropic is coinciding with reported OpenAI discounts aimed at business customers using rival AI services.
Thorsten Meyer AI
List prices · Jun–Oct 2026

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.

−67%OpenAI Sol output price, June ($30) to September ($10)
−92%OpenAI Luna output price, June ($6) to September ($0.50)
−20%Anthropic Opus output price, Opus 5 ($25) to Opus 5.5 ($20)
−47%average price per token on Vercel’s gateway, December to September

OpenAI · output price per 1M tokens

List prices, standard tier

  • GPT-5.6 Solpreview, 26 Jun$30
  • GPT-5.6 Solpromo, 21 Aug$20
  • GPT-6 / 6.1 Sol22 + 29 Sep$10
  • GPT-5.6 Lunapreview, 26 Jun$6
  • GPT-5.6 Lunacut, 30 Jul$1.20
  • GPT-6 Luna22 Sep$0.50

Anthropic · output price per 1M tokens

List prices, standard tier

  • Claude Opus 524 Jul$25
  • Claude Opus 5.522 Sep$20
  • Sonnet 5, plannedstandard price$15
  • Sonnet 5, permanent10 Aug$10

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

  • GPT-5.6 SolJune$0.50
  • GPT-6.1 Sol29 Sep$0.10
  • Claude Opus 5July$0.50
  • Claude Opus 5.522 Sep$0.20
  • Claude Fable 5June$1.00
  • Claude Fable 5.1September$0.25

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 Rates, More Contract Leverage

The reductions give companies buying large volumes of AI services more room to negotiate and make it easier to compare suppliers. A lower token price can affect the economics of coding assistants and other products that repeatedly send text to a model. But published rates alone do not establish what a customer will pay: volume discounts, model choice, caching, and contract terms all affect the bill.

The competition is also affecting where customers direct work. Rippling CEO Parker Conrad said OpenAI’s share of coding work at his company rose from 16% in May to 52%, while Anthropic’s share fell from 57% to 18%. His account links the shift to cost, but it is a statement from one customer, not evidence of a broad industry-wide migration.

For smaller buyers and individual users, the main issue may be the terms attached to a plan rather than a headline token rate. The source reports that allowances and credit rules have changed at both providers. A subscription’s monthly price can stay the same even when its included usage or access conditions change, making direct comparisons harder.

Products Worth Considering

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

LG UltraGear 27-inch QHD OLED Gaming Monitor, 240Hz, 0.03ms, G-SYNC Compatible, FreeSync Premium Pro, HDR True Black 400

As an affiliate, we earn on qualifying purchases.

How Prices Changed Since June

The price sequence began with model announcements in June and continued through September. Anthropic launched Sonnet 5 at an introductory rate of $2 per million input tokens and $10 per million output tokens, then made that price permanent. OpenAI cut Luna pricing in July, temporarily reduced Sol rates in August, and introduced lower prices for successor models in September. Woolf’s timeline, dated October 25, tracks those moves; the source says its figures were checked against vendor announcements and pricing pages the following day.

Some cuts apply to new model generations, so they do not prove that an equivalent service has simply become cheaper. Performance may differ by task, and buyers need to test models against their own work. Cache pricing is another factor: the source notes that agentic coding systems may reuse substantial amounts of prior context, so cache costs can matter alongside standard input and output rates. The precise share of spending attributable to cached tokens is Woolf’s estimate, not a vendor-wide audited measure.

Published prices are only part of the commercial picture. The source also reports that Anthropic added capped, expiring API credits to its Max and Team plans, and that its Enterprise plan requires a 20-seat minimum. It says OpenAI reduced usage included with new $200 Pro subscriptions while keeping the price unchanged. Those account terms can influence a buyer’s effective cost even when list prices fall.

What the Savings Figures Cover

The reported business offers do not establish standard discounts available to every customer. Contract details, eligibility, duration and usage conditions have not been fully disclosed in the source material. The Information’s report is paywalled, and the account summarized here relies on its public summary and subsequent coverage by The Decoder and all-ai.de.

It is also unclear how much of the reported savings will hold once customers account for model performance, usage patterns and changing plan limits. Some price comparisons are between successive models, whose results may not be equivalent for a particular task. Replit’s exact current mix of model providers is not established: a source cited by TechCrunch said the partnership remained in place, despite Masad’s comments about a preferred partner.

The source’s account of Anthropic’s seat minimums and OpenAI’s Pro usage changes does not supply full plan terms or a complete date-by-date record. Buyers should check current vendor documentation and their own contracts before treating any reported rate or allowance as applicable to them.

What Buyers Should Check Next

Companies will need to compare total costs under their own workloads, rather than rely on headline token prices or a single customer’s savings estimate. That means checking input, output and cache rates alongside committed spend, credits, seat minimums, expiration rules, usage caps and renewal terms. Testing successor models on representative tasks can show whether a lower rate delivers comparable results for a given use.

Further vendor pricing updates and customer disclosures may clarify how widespread negotiated discounts are and whether they persist. Until those details emerge, the confirmed picture is a series of list-price reductions and several reported commercial offers—not a uniform discount across the market.

Key Questions

Have OpenAI and Anthropic cut their published AI prices?

Yes. The source material reports price reductions across several model releases from June through September 2026. For example, OpenAI’s Sol pricing moved from $5 input and $30 output per million tokens in June to $2 and $10 for GPT-6.1 Sol in September. These comparisons involve different model versions.

Does the 80% to 90% figure mean OpenAI cut list prices by that much?

No. DataCamp CEO Jonathan Cornelissen described expected savings against his company’s current setup. That is not a published discount of 80% to 90% off OpenAI’s standard prices.

Are companies switching from Anthropic to OpenAI?

Some reported customer use has shifted. Rippling CEO Parker Conrad said OpenAI’s share of coding work at his company increased from 16% in May to 52%, while Anthropic’s share fell from 57% to 18%. The figures concern Rippling and do not establish a market-wide trend.

Why can a lower token price fail to reduce a bill by the same amount?

Effective cost also depends on the model and task, token usage, cache rates, negotiated contract pricing, subscription allowances, credits and seat requirements. A new model’s price is not a like-for-like comparison unless it performs similarly on the buyer’s workload.

What should AI customers monitor now?

Check current pricing and plan terms, including usage caps, included allowances, credit expiration, seat minimums, renewal conditions and rates for cached tokens. Businesses can also test more than one provider on representative work before committing to a contract.

Source: ThorstenMeyerAI.com

You May Also Like

Fable 5 – Median Thinking Declined In August

Analysis shows a decline in median player thinking levels for Fable 5 during August, indicating possible shifts in player engagement and game perception.

AI prompt audit log for marketing agencies

Small marketing agencies are testing a new AI prompt and output logging system to improve review and approval processes for client work.

Grok Vs ChatGPT: 3 Things To Know About Their 2026 Output Costs

A 2026 headline lists Grok at $6 and ChatGPT at $10 per million outputs, but the available material does not define the unit or support the prices.

Music And Live Events Signal Monitor: Ella Langley Tulsa

IdeaNavigator AI proposes testing a live-events monitor around “ella langley tulsa,” but provides no evidence of a confirmed show or booking.