Anthropic Launches Claude Haiku 5.5 With Major Price Cuts
Table of Contents
Anthropic has launched Claude Haiku 5.5, its newest lightweight AI model, with significantly lower API prices and improved performance for high-volume workloads. Released on October 7, the model is the third addition to the company’s Claude 5.5 family within a month.
The launch puts cost efficiency at the center of Anthropic’s latest AI strategy. The company says Claude Haiku 5.5 costs about 75% less to run on average than Haiku 4.5, while some customers can see price reductions of up to 90%.
Claude Haiku 5.5 Brings Major Price Reductions
For requests containing up to 100,000 tokens, Claude Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens. By comparison, Haiku 4.5 charged $1 for input and $5 for output.
Requests exceeding 100,000 tokens are priced at $0.50 per million input tokens and $2.50 per million output tokens. That represents a 50% reduction compared with the previous model’s rates.
Anthropic says roughly 90% of Haiku 4.5 requests were already within the cheaper token range. The company’s 75% average cost reduction also accounts for a revised tokeniser that can use slightly more tokens to complete the same task.
The new pricing makes Claude Haiku 5.5 particularly attractive for applications that process large volumes of requests.
Designed for High-Volume AI Work
Anthropic is positioning Claude Haiku 5.5 for practical tasks that need speed, scale, and controlled costs rather than the maximum possible reasoning capability.
The company highlights use cases such as:
- Text classification and summarisation
- Information extraction
- Customer support
- Browser-based computer tasks
- Repetitive agent workflows
- High-volume developer applications
Claude Haiku 5.5 can also work as a sub-agent alongside Anthropic’s more capable models. In this setup, a larger model can handle planning and complex decisions while Haiku performs clearly defined tasks quickly.
The model is also Anthropic’s first Haiku release to include effort controls. Developers can adjust the amount of reasoning used, creating a trade-off between performance and operating costs. The default setting is medium.
| Model | Input Price | Output Price | Positioning |
| Claude Haiku 5.5 | $0.10/M | $0.50/M | Fast, low-cost workloads |
| Claude Haiku 4.5 | $1/M | $5/M | Previous Haiku model |
| Claude Haiku 5.5, long prompts | $0.50/M | $2.50/M | Over 100K tokens |
| Claude Sonnet 5.5 cached reads | $0.10/M | — | Lower agent costs |
Claude Haiku 5.5 Shows Stronger Benchmark Results
Lower pricing is not the only change. Anthropic says Claude Haiku 5.5 has made substantial progress on computer-use and coding-related evaluations.
On the offline portion of the OSWorld 2.1 computer-use benchmark, Claude Haiku 5.5 scored 72.4%, compared with 15.7% for Haiku 4.5. Anthropic’s figures also place it ahead of OpenAI’s GPT-6 Luna, which scored 48.9% on the same test.
On Terminal-Bench 4.0, Haiku 5.5 achieved 39.2%, compared with 16.4% for GPT-6 Luna. However, it remained behind Sonnet 5.5, which scored 70.6%.
These results suggest Anthropic is trying to make its smaller model capable enough for increasingly sophisticated automated workflows without requiring developers to pay premium-model prices.
The model is available through Anthropic’s Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure.
Anthropic Cuts More AI Costs
The Haiku release comes alongside additional pricing changes across Anthropic’s ecosystem. The company has reduced the price of cached reads on Sonnet 5.5 by half, bringing the cost to $0.10 per million tokens.
Anthropic estimates that this change can make many agentic tasks around 20% cheaper. The company is also introducing monthly API credits for Max and Team subscribers.
Max 5x customers will receive $100 in monthly API credits, while Max 20x users receive $200. Team plans can receive up to $500 in shared credits.
These changes indicate that Anthropic is focusing heavily on the economics of AI development as companies increasingly build agents that make repeated model calls.
FAQs About Claude Haiku 5.5
What is Claude Haiku 5.5?
Claude Haiku 5.5 is Anthropic’s latest lightweight AI model, designed for fast, scalable, and cost-efficient workloads.
How much does Claude Haiku 5.5 cost?
For requests up to 100,000 tokens, pricing is $0.10 per million input tokens and $0.50 per million output tokens.
What can Claude Haiku 5.5 be used for?
It is designed for classification, summarisation, information extraction, customer support, browser tasks, and sub-agent workflows.
Is Claude Haiku 5.5 available on cloud platforms?
Yes. It is available through Anthropic, AWS, Google Cloud, and Microsoft Azure.
Does Haiku 5.5 have reasoning controls?
Yes. Developers can use effort controls to balance reasoning quality and operating costs.
Conclusion
Claude Haiku 5.5 strengthens Anthropic’s position in the fast-growing market for affordable AI models. Its lower prices, improved computer-use performance, and flexible reasoning controls make it suited to developers running large numbers of automated tasks.
As AI agents become more common, reducing the cost of repeated model calls could become as important as improving raw intelligence. With Claude Haiku 5.5, Anthropic is betting that powerful AI can become both more capable and significantly cheaper to operate. That balance could make Haiku 5.5 especially competitive for developers.
