Haiku 5.5
A lightweight model in the Anthropic Claude 5.5 series, designed for high-volume, cost-sensitive application scenarios. It represents the most economical option within the Claude model family.
Core Facts
Compared to Claude Haiku 4.5, Haiku 5.5 reduces operating costs by an average of approximately 75%.
Once the context exceeds 100,000 tokens, the price of Haiku 5.5 increases 5x directly: output goes from $0.50 to $2.50 per million tokens, and input goes from $0.10 to $0.50.
Haiku 5.5's input price has dropped to $0.10 per million tokens
Timeline (last 90 days)
Anthropic charges standard rates for the full 1-million-token window of current models (except Haiku 5.5 above 100,000 tokens), while OpenAI has set separate and higher long-context rates for Astra, Sol, and Luna.
Anthropic made a similar model split the same month: Haiku 5.5 targeting high-throughput scenarios and Opus 5.5 targeting frontier reasoning.
Putting Explore on Haiku 5.5 can reduce the main model's token consumption and speed up responses for exploration-type tasks.
The original author verified through multiple rounds of testing the feasibility of adjusting the model resolution order to force Explore to use Haiku 5.5.
Claude Code's Explore subagent by default runs following the main model, and merely setting the CLAUDE_CODE_SUBAGENT_MODEL environment variable cannot switch it to Haiku 5.5.
Opus 5.5 was released on September 22, Sonnet 5.5 was released on September 28, and Haiku 5.5 was released on October 7.
Claude Sonnet 5.5 sits between Haiku 5.5 and Opus 5.5 in Anthropic's product line, positioned as the primary enterprise-grade workhorse.
Haiku 5.5 is designed for fast, repetitive workloads such as summarization, compression, database queries, and classification, as well as latency-sensitive scenarios such as real-time customer service and browser operations.
There is often a gap between benchmark results and real-world business experience, and actual performance depends on the task type, reasoning settings, and calling method.
Haiku 5.5 supports a context window of 1 million tokens.
38 more timeline events
All Facts (20)
Compared to Claude Haiku 4.5, Haiku 5.5 reduces operating costs by an average of approximately 75%.
90%VerifiedOnce the context exceeds 100,000 tokens, the price of Haiku 5.5 increases 5x directly: output goes from $0.50 to $2.50 per million tokens, and input goes from $0.10 to $0.50.
70%VerifiedHaiku 5.5's input price has dropped to $0.10 per million tokens
70%VerifiedHaiku 5.5 is the first Haiku model with an adjustable effort setting.
70%VerifiedHaiku 5.5 is the first Haiku model to support adjustable effort, allowing developers to select the amount of reasoning resources invested based on task difficulty.
65%VerifiedHaiku 5.5 can work as a sub-agent alongside Claude Opus 5.5 and Sonnet 5.5 to complete programming tasks.
65%VerifiedAnthropic has released Haiku 5.5, describing it as the cheapest, fastest, and most capable small model the company has ever released.
65%UnverifiedAnthropic charges standard rates for the full 1-million-token window of current models (except Haiku 5.5 above 100,000 tokens), while OpenAI has set separate and higher long-context rates for Astra, Sol, and Luna.
50%UnverifiedClaude Code's Explore subagent by default runs following the main model, and merely setting the CLAUDE_CODE_SUBAGENT_MODEL environment variable cannot switch it to Haiku 5.5.
50%UnverifiedPutting Explore on Haiku 5.5 can reduce the main model's token consumption and speed up responses for exploration-type tasks.
50%UnverifiedClaude Sonnet 5.5 sits between Haiku 5.5 and Opus 5.5 in Anthropic's product line, positioned as the primary enterprise-grade workhorse.
50%UnverifiedThe Claude 5.5 series includes three models—Opus 5.5, Sonnet 5.5, and Haiku 5.5—positioned as flagship, general-purpose, and lightweight respectively.
50%UnverifiedHaiku 5.5 is designed for fast, repetitive workloads such as summarization, compression, database queries, and classification, as well as latency-sensitive scenarios such as real-time customer service and browser operations.
50%UnverifiedHaiku 5.5 supports a context window of 1 million tokens.
50%UnverifiedHaiku 5.5's score on Terminal Bench 4.0 rose from 0% in the previous generation to 39.2%.
50%UnverifiedAnthropic states that Haiku 5.5 outperforms GPT-6 Luna on certain metrics including GDPVal and Agentic Coding.
50%UnverifiedThere is often a gap between benchmark results and real-world business experience, and actual performance depends on the task type, reasoning settings, and calling method.
50%UnverifiedHaiku 5.5 achieved a score of 1620 on the relevant coding benchmark, far higher than Haiku 4.5's score of 735.
50%UnverifiedThe company explicitly states that complex agentic coding tasks should use Sonnet or Opus rather than Haiku 5.5.
50%UnverifiedAll four models have a context length of 1 million tokens.
50%