What To Read Before Signing Up For OpenAI Or Anthropic

🔍 Read the full analysis: What To Read Before Signing Up For OpenAI Or Anthropic on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

• Business-only prices and quantity discounts
• Tax-exempt purchasing
• Multiple users, one account, clear invoices

Create a free Business account →Free to register

As an affiliate, we earn on qualifying purchases.Hide

OpenAI and Anthropic have lowered prices for several AI models in recent months, while reported customer deals and changes to subscription terms have made the fine print a central issue for buyers. A review of vendor pricing pages, retrieved October 26, and reporting by The Information point to lower costs in some cases, but the figures do not amount to a universal discount: actual savings depend on usage, negotiated terms and the model selected.

OpenAI’s pricing for its Sol model line 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, according to the source’s review of vendor pricing. Its smaller Luna model moved from $1 and $6 per million tokens to $0.10 and $0.50. Anthropic also reduced prices across releases: Opus went from $5/$25 per million input/output tokens for Opus 5 in July to $4/$20 for Opus 5.5, while cache reads for Opus 5.5 were listed at a lower price.

Those comparisons involve successor models, not necessarily the same model becoming cheaper. Anthropic has also said its newer models cost less on typical workloads, but those workload estimates are vendor claims. Buyers need to test whether a newer model performs adequately on their own tasks and to calculate costs using their actual pattern of prompts, outputs and cached tokens.

The Information reported October 24 that OpenAI is offering discounts to attract customers using Anthropic. The specific examples in the source account vary: a person cited by The Information said Replit received an offer of more than 30% off a contract valued at over $100 million annually. DataCamp CEO Jonathan Cornelissen described expected total cost reductions of 80% to 90% under OpenAI’s pricing. That figure is an estimate against DataCamp’s existing setup, not a published 90% reduction from OpenAI’s list price.

Contract Terms Can Change Total Cost

Lower token prices can make AI more affordable, but they do not alone establish what a company will pay. Seat counts, usage allowances, credit expiry and minimum commitments can change a contract’s value, even when the monthly fee or headline rate looks attractive. The source says Anthropic added capped, expiring API credits to its Max and Team plans and requires 20 seats for Enterprise. That minimum could put the plan out of reach for smaller organizations.

Subscription terms also need checking over time. The source reports that OpenAI reduced the usage included with new $200 Pro subscriptions without lowering the price. For individual subscribers, this means a stable monthly fee may not guarantee a stable allowance. For businesses, negotiating multiple vendors and testing workloads can help them compare the cost of delivering the same work rather than comparing token prices in isolation.

There is evidence that at least some customers are shifting work. 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%. That is one company’s reported usage mix, not evidence of a market-wide shift or proof that price alone caused it.

How the Price Changes Accumulated

The source traces multiple price moves from June through September 2026. OpenAI reduced Luna pricing in July and cut Sol pricing temporarily in August before introducing GPT-6 models at lower rates in September. Anthropic set Sonnet 5’s introductory price as its permanent price in August and released Opus 5.5 in September. The source says it checked the sequence against vendor announcements and pricing pages retrieved October 26.

For some AI workloads, the cost of reading cached tokens can matter alongside standard input and output rates. The source cites data scientist Max Woolf’s estimate that cached tokens account for 96% to 99% of token spending in agentic coding. Anthropic has said its Fable 5.1 model costs around 25% less on typical workloads and up to around 45% less on highly agentic workloads; these are Anthropic’s estimates, not independently established savings for every customer.

Customer reports also need careful reading. 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 OpenAI was its preferred partner. A person familiar with the partnership told TechCrunch it was still in place. These statements describe a changing relationship, not a clear public account of every contract term.

“I think ChatGPT is better now.”

— Amjad Masad, Replit CEO, speaking at TechCrunch Disrupt

Savings Depend on Workload and Terms

Public information does not establish a standard discount available to all OpenAI or Anthropic customers. The Information’s article is paywalled, and the source account relies on its public summary and other reports for some deal details. The reported offers may reflect private negotiations, contract size and customer circumstances.

It is also unclear how much of each reported cost reduction comes from lower rates, a change in model mix, usage patterns or other contract terms. The claim that a newer model costs less does not establish that it will perform equally well on every task. Buyers should distinguish a vendor’s benchmark or workload estimate from their own measured results.

Some plan details may change, and the source does not provide a complete comparison of every current subscription’s limits, exclusions or renewal provisions. The Replit accounts also leave the precise status and scope of the partnership unclear.

Verify Terms Before Committing

Before signing or renewing, buyers can request a written account of unit prices, included usage, overage charges, credit caps and expiry dates, along with any seat minimums and renewal rules. They should compare the vendors using representative workloads and record the quality and cost of the results. Where feasible, testing more than one model can show whether a lower-priced option meets operational needs.

For current subscribers, the next step is to check whether allowances or included usage have changed, even if the monthly price has not. Companies considering a large commitment should ask vendors to specify how discounts apply over the contract term and what happens when usage exceeds the allowance. The available reporting does not say how broadly the reported offers will spread; pricing pages and plan terms remain subject to change.

Key Questions

Does OpenAI offer every customer a 90% discount?

No. The 80% to 90% figure is DataCamp CEO Jonathan Cornelissen’s estimate of expected total savings against the company’s existing setup, as reported by The Information. It is not a general discount off OpenAI’s public prices.

Which OpenAI prices fell?

According to the source’s review of vendor pricing, Sol moved from $5/$30 to $2/$10 per million input/output tokens between June and September, while Luna moved from $1/$6 to $0.10/$0.50. These comparisons involve newer model releases.

What should a business check before signing?

Check token rates and expected usage, as well as seat minimums, included allowances, overage charges, credit caps and expiry dates. Test models on representative work before relying on vendor savings estimates.

Are the new models guaranteed to perform as well as their predecessors?

No such guarantee is established by the source. A lower price for a successor model does not prove equivalent performance for a specific task; buyers should run their own evaluations.

Source: ThorstenMeyerAI.com