Anthropic releases Claude Haiku 5.5, a small model for high-volume work

Anthropic announced Claude Haiku 5.5 on 7 October 2026. The company's announcement says the model targets repetitive, large-volume work where cost matters, such as summaries, context compaction, database queries and classification. Haiku 5.5 is the first Haiku-class model with an adjustable effort setting from Low to Max, so users can choose between cost and intelligence.

• For prompts up to 100 thousand tokens, pricing is 0.10 dollars per million input tokens and 0.50 dollars per million output tokens; Anthropic says it costs about 75 percent less to run than Haiku 4.5 on average.
• In Anthropic's published results it scored 72.4 percent on the offline subset of OSWorld 2.1, a computer-use test, compared with 15.7 percent for Haiku 4.5.
• For complex agentic coding, the company still points users to its larger Sonnet 5.5 and Opus 5.5 models.
• In the same announcement, the cache-read price of Sonnet 5.5 was cut by 50 percent, from 0.20 to 0.10 dollars per million tokens.

Why it matters

When tasks like sorting customer messages, summarizing requests or pulling reports from a database run thousands of times a day, model cost can become decisive. For companies in Türkiye that serve customers over WhatsApp or the web, low unit prices and speed can be among the main factors in whether automation fits the budget. The announcement does not state a context window for the model; before choosing, it is worth comparing Haiku 5.5 with larger models on your own sample data.