Claude Haiku 5.5 released with 1/10th price and 1M token context
Claude released its new lightest model, Claude Haiku 5.5, on October 7th. The model is priced at 1/10th the cost of the previous Haiku 4.5, with input costs dropping to 5 cents per 1 million tokens for texts under 100,000 tokens. The model now supports a 1 million token context window, up from 200,000, and offers five levels of reasoning depth.
What you can do with it
AI at Work →Draft 30 follow-up emails in seconds, right inside Claude.ai
What you get You can draft student follow-ups.
Generates personalized follow-up emails for multiple recipients at a very low cost.
- Who for
- marketers, sales and founders
- Cost
- free tier
- Effort
- minutes
Use it for
- Draft student follow-ups
- Summarize feedback
- Sort questions
How to set it up · From the article
- Open Claude.ai and select Haiku 5.5 from the model list.
- Ask it to write follow-up emails for a specific group.
- Read each email and delete any sentences that are factually incorrect.
Watch out Model may invent plausible but false details; always review before sending
Key points from the news
- Claude Haiku 5.5 costs 5 cents per 1 million input tokens for texts under 100,000 tokens.
- The model supports a 1 million token context window and five reasoning depth levels.
- Testing showed Haiku 5.5 invented fictional details in emails, while Sonnet 5.5 used clear placeholders.
The author tested the model by generating 30 follow-up emails, which cost 0.44 yen in total. While Haiku 5.5 correctly identified a date error in the prompt, it also invented plausible but false details about student behavior. In a comparison for a diagnostic tool, Haiku 5.5 was faster but used fictional placeholders like "Company A," whereas Sonnet 5.5 used clear brackets like "[Name]." The author decided to use Haiku for high-volume internal drafts and Sonnet for text sent directly to recipients to avoid errors.
The model is available on all Claude.ai plans, including the free tier, and in Claude Code. The author advises choosing models based on who will read the output: Haiku for cheap, high-volume tasks that require human review, Sonnet for client-facing text, and Opus for complex strategic thinking.
Model pages: Sonnet 5.5 → · Opus 5.5 →
The story so far
5 episodes →- Claude Haiku 5.5 released with 1/10th price and 1M token contextthis story
I asked the cheapest Claude to write thank-you emails for 30 people, and it cost 0.44 yen
note.com · 24 October 2026
Loading the full article…
This text was published by note.com and written by 工藤 晶｜講師のAIパートナーづくり. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Marketing & Small Business
All →- Meta launches Muse AI agent for task automation in Japan and US · 10 src
- Slack adds AI agents and Salesforce CRM tools for small business · 1 src
- Anthropic launches Claude Duet, a proactive AI agent app for Pro and Max users · 26 src
- Canva launches Canva Craft image generation model for small businesses · 2 src
- Google announces increase in links to original pages in AI answers · 1 src
Comments
via GitHub Discussions