# Anthropic releases Claude Haiku 5.5 with 1M context and lower costs

Digest AI · Generative AI & Models · published 2026-10-19T16:00:00Z

Canonical: https://digestai.news/story/anthropic-releases-claude-haiku-5-5-with-1m-context-and-lower-costs

## Summary

Anthropic released Claude Haiku 5.5 on October 7, 2026, positioning it as its fastest and most cost-effective small model for high-volume tasks. The new model features a 1M token context window and adaptive thinking. While list prices for prompts up to 100,000 tokens drop to $0.10 per million input tokens and $0.50 per million output tokens, Anthropic notes that a newer tokenizer increases token counts by roughly 30%. This results in an average cost saving of around 75% compared to previous versions, rather than the 90% suggested by list prices alone.

The release includes breaking changes for developers: sending temperature, topp, or topk parameters now returns an error, as does manual thinking with budgettokens. Adaptive thinking is enabled by default, meaning thinking tokens count toward the maxtokens limit. Safety classifiers are stricter, potentially declining requests with a "refusal" stop reason, which requires code updates to handle properly.

The article provides Python code examples for using Haiku 5.5 in classification, extraction, and sub-agent workflows. It highlights that while Haiku 5.5 excels at narrow, repetitive jobs like summarization and routing, larger models like Sonnet 5.5 or Opus 5.5 remain better suited for complex agentic coding. Companies like Cognition and AlphaSense have already integrated the model, reporting lower latency and improved performance on specific document tasks.

## Key points

- Claude Haiku 5.5 costs $0.10 per million input tokens for prompts up to 100,000 tokens.
- A new tokenizer increases token counts by roughly 30%, reducing average savings to about 75%.
- Sending temperature, topp, or topk parameters now returns an error in the API.

## Why it matters

Haiku 5.5 significantly reduces the cost of high-volume AI tasks like classification and extraction. Its 1M context window and speed make it ideal for sub-agent architectures, allowing developers to build cheaper, faster workflows while reserving expensive models for complex reasoning.

## Sources

1. [Claude Haiku 5.5: What’s New and How to Use It for Fast, Cheap AI Workflows](https://kdnuggets.com/claude-haiku-5-5-whats-new-and-how-to-use-it-for-fast-cheap-ai-workflows) (KDnuggets, 2026-10-19)

Part of the developing story: [AI Coding Wars GPT vs Claude Showdown](https://digestai.news/thread/anthropic-releases-free-crash-course-on-building-ai-agents-with-opus-5-5) (3 stories)

## Cite

Digest AI, "Anthropic releases Claude Haiku 5.5 with 1M context and lower costs", 19 October 2026, https://digestai.news/story/anthropic-releases-claude-haiku-5-5-with-1m-context-and-lower-costs

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/anthropic-releases-claude-haiku-5-5-with-1m-context-and-lower-costs.json