Claude Haiku 5.5: Anthropic launches an AI 4 times cheaper to compete with ChatGPT

Anthropic wants to make AI accessible to every budget. This Wednesday, October 7, the American company unveiled Claude Haiku 5.5, a new version of its smallest generative AI model.Faster and more performant than its predecessor, the AI is nearly four times cheaper than the previous iteration, Haiku 4.5.

Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released.

On average, it costs around 75% less to run than Claude Haiku 4.5. https://t.co/jm06cZJfkV

— Claude (@claudeai) October 7, 2026

Anthropic presents it as “the cheapest, fastest, and most capable small model” it has ever created. For reference, Anthropic divides Claude into three distinct families: Haiku, the smallest and cheapest; Sonnet, the mid-range model; and Opus, the most powerful.

Read also: ChatGPT vs Claude – OpenAI and Anthropic are rolling out cheaper, more effective AIs… instead of slowing down

An AI for everyday tasks

Haiku 5.5 is not designed for every use case. The model is intended for simple, repetitive tasks—those that users will perform millions of times. These include summarizing a text, classifying information, querying a database, or browsing the web automatically. Haiku 5.5 can also support a more powerful AI as part of a complex task. For the most challenging programming projects, Anthropic still recommends turning to Sonnet 5.5 or Opus 5.5.

In terms of pricing, the reduction is dramatic.Haiku 5.5 costs businesses 90% less for short requests and 50% less for longer ones. With the new version of Claude, Anthropic is mirroring the offering of its biggest rival, OpenAI. The prices are indeed exactly the same as those of GPT-6 Luna, the economical model that powers ChatGPT. As a result, Anthropic is also reducing the price of certain operations performed with Sonnet 5.5.

Also read: To train Claude, Anthropic wants your voice

Other new features of the model

More affordable, Haiku 5.5 is also far more capable than its predecessor, Anthropic says. It stands out especially for its improved ability to control a computer. Another novelty for a Haiku model, users can also adjust the AI’s level of effort, choosing between savings and intelligence. On the security front, Anthropic says its model exhibits far fewer undesirable behaviors than its predecessor. Its safeguards notably block penetration tests, which involve simulating an attack to uncover vulnerabilities in a system.

As you may have gathered from the recent announcements by industry giants, the AI race has moved away from performance and focused instead on pricing.As calls to slow down AI research multiply, the goal is no longer to offer the most powerful model, but to make available cost-effective solutions, designed for adoption by the widest possible number.

👉🏻 Follow tech news in real time: add 01net to your sources on Google, and subscribe to our WhatsApp channel.