New · See every ThinkChat feature in action. Take a look →
ThinkChat
Start free trial

Knowledge AI

AI is getting cheaper to run. What Claude Haiku 5.5’s prices mean for small businesses

AI 3 min read

On 7 October 2026, Anthropic released Claude Haiku 5.5, its newest small AI model. Anthropic calls it “the cheapest, fastest, and most capable small model we’ve ever released,” and says that on average it “costs around 75% less to run” than the older Haiku 4.5.

Anthropic also says that, because it is its fastest model to date, Haiku 5.5 “works especially well for speed-sensitive tasks like live customer support.”

If you run a small business, you probably won’t see these prices directly unless you build your own AI tools. But they are part of what it costs to run an AI Chat Agent for customer service behind the scenes.

What Anthropic announced

  • New prices. For prompts up to 100,000 tokens, Haiku 5.5 costs US$0.10 per million input tokens and US$0.50 per million output tokens. Haiku 4.5 costs US$1 per million input tokens and US$5 per million output tokens. (Anthropic describes tokens as the smallest units a language model works with. A token can be a word, part of a word, or a single character.)
  • Longer prompts cost more. For prompts over 100,000 tokens, Haiku 5.5 costs US$0.50 per million input tokens and US$2.50 per million output tokens.
  • Why “around 75%” and not 90%. Anthropic says Haiku 5.5 is priced 90% lower than Haiku 4.5 for requests up to 100,000 tokens and 50% lower for longer ones, and that 90% of Haiku 4.5 requests were the shorter kind. Haiku 5.5 also uses slightly more tokens per task, so Anthropic puts the average saving at around 75%.
  • Extra discounts. Anthropic says you can save up to 90% with prompt caching (reusing parts of a prompt that were already processed) and 50% with batch processing (jobs that don’t need an instant answer).
  • Bigger models still have a job. Anthropic says its larger models, Sonnet 5.5 and Opus 5.5, “remain better choices for complex agentic coding tasks,” and that Haiku 5.5 “is best suited to more narrowly scoped tasks.”

What this means for a small business

*The tips below are our suggestions, drawing on Anthropic’s published information above. They are not Anthropic’s advice.*

1. Answering routine questions is getting cheaper. Questions about opening hours, prices, what’s included and how to book are short and repetitive. That is the kind of high-volume work Anthropic says Haiku 5.5 is designed for. In our view, the AI model itself is likely to become a smaller part of the cost.

2. Cheap doesn’t mean right. A cheap model that gives a customer the wrong price, or makes up a policy, can cost you far more than the model does. What still matters is whether your AI answers from information you approved, stays honest when it doesn’t know, and can hand over to a person.

3. Use the right size for the job. Anthropic positions Haiku 5.5 for narrower, high-volume tasks and its larger models for more complex work. For customer service, we suggest letting quick, routine answers run on AI, while complaints, refunds and unusual requests go to a person.

4. Ask your vendor how it bills. Many AI Chat Agents for customer service charge a monthly plan rather than per token, so check how yours bills. Ask what you pay per month, what’s included, and what happens when your chat volume grows.

Where ThinkChat fits

ThinkChat is an AI Chat Agent and multi-channel AI customer service platform. WhatsApp is the main channel in Malaysia, and ThinkChat also works on Instagram, Messenger, Telegram, website chat and more. However cheap AI gets, we think a good AI Chat Agent answers from information you approved, stays honest when it doesn’t know, and can hand over to a person.

Soft next step

Want an AI Chat Agent to answer your customers 24/7? Talk to Lyia on WhatsApp: +60 11-5515 9317.

Sources

Checked 23 October 2026.

← All articles