Anthropic launches cheaper Claude Haiku 5.5 to complete its lineup before IPO

On October 7, Anthropic released Claude Haiku 5.5, adding the last piece to its trio lineup of products, which includes Sonnet 5.5 and Opus 5.5.

The new model is aimed at assisting users with their daily work processes, as organizations are searching for ways to lower the costs of using AI. The rollout is also occurring at a time when Anthropic is getting ready for its initial public offering.

Priced to run at scale

According to Anthropic, Haiku 5.5 is approximately 75% cheaper than Haiku 4.5 on average. For requests of up to 100,000 tokens, the pricing stands at $0.10 for every million input tokens and $0.50 for every million output tokens.

This model works with simple tasks including summarization, classification, database inquiries, and actions performed by artificial intelligence agents.

Moreover, Anthropic has managed to decrease cache-read costs in case of Sonnet 5.5. In its launch testimonials, Asana mentioned that they had reduced latency by over 30%, while Box showed an improvement of 11 points over version 4.5.

A market growing 63% a year

The numbers indicate that businesses are rapidly increasing their spending on AI.

According to Gartner, spending on AI models and platforms is estimated to increase by more than 60%, from $39 billion in 2025 to $64 billion this year.

As AI spending grows, companies are becoming more selective about where they invest their money. A September study by BCG found that nearly half of businesses are already observing returns in terms of value coming from AI. This increases the demand for companies to charge reasonable prices, as well as improve their results.

Why the biggest model isn’t always the pick

The strongest AI model doesn’t necessarily mean it is the best for every task.

In an analysis conducted by the World Economic Forum on August 27 involving 44 supplier invoices, a mid-level AI solution was able to handle all invoices while a more powerful solution was able to solve only 91% of them.

The findings suggest that a model’s suitability for the task is more important than its ranking. As Cryptopolitan reported, competition is increasingly shifting beyond benchmark scores to the cost of completing a task.

Gaining on OpenAI, but loyalty is thin

Anthropic is establishing itself among the business customers.

The summer report by Ramp, which examined the expenditure of more than 70,000 businesses, indicates that Anthropic has surpassed OpenAI in terms of adoption in the American market.

However, maintaining that lead may prove to be difficult. This is because more than half (52%) of businesses using Anthropic or OpenAI are customers of both providers.

Ramp’s separate research shows that the percentage of switching providers in September reached a record of 8%. This suggests that businesses appear increasingly comfortable moving between competing AI services.

The question before the IPO

Lower prices could help Anthropic attract more customers and encourage existing ones to use its models more often. That strategy could strengthen its position ahead of a possible IPO, but the company still needs to show that growing demand can translate into sustainable profits.

Anthropic is targeting a public listing after the November 2026 US midterm elections, although no firm date has been announced. A September 28 Reuters report said the offering could value the company at more than $2 trillion. Its prospectus also revealed the financial pressures behind that valuation. Revenue surged twelvefold to nearly $4.6 billion in 2025, but Anthropic recorded an operating loss of $8.06 billion and disclosed $518 billion in future cloud, computing and infrastructure commitments.

Those costs reflect the enormous resources needed to develop and operate advanced AI systems. They also raise the stakes for Anthropic’s lower-cost model strategy. While cheaper models could encourage wider adoption, higher usage must eventually generate enough revenue to cover the company’s growing expenses.

So now, Anthropic has an important problem on its hands as it prepares for an IPO: will it be able achieve sustainable profits while continuing to invest heavily in AI infrastructure?