• Home
  • Search
  • AI Services
  • Models
  • Writing
  • Multimedia Creation
  • Translation & Others

Anthropic Claude Haiku 5.5

Anthropic Claude Haiku 5.5, the new-generation lightweight model: the fastest and most capable small Claude model, built for high-volume classification, summarization, and extraction.

Anthropic Claude Haiku 5.5, released in October 2026, is the new-generation lightweight model, positioned as the fastest, most economical, and most capable small model Anthropic has released, built for high-volume, latency-sensitive work such as classification, summarization, extraction, database queries, and customer support, and well suited as a subagent for Opus 5.5 and Sonnet 5.5. Against Haiku 4.5, agentic coding on Terminal-Bench jumps from 0 to about 39 percent, computer use rises from about 16 to over 72 percent, and knowledge work approaches mid-tier models; it is the fastest model in the Claude family. Its knowledge cutoff is June 2026, and it supports very long context for processing many documents at once. The interface offers a reasoning-effort dropdown (five levels, default medium), an option to show the thinking process, web search, image and document upload, and continuous conversations. Use it for high-volume lightweight workloads; switch to Sonnet 5.5 or Opus 5.5 for deep reasoning or long-form writing. If you need to search the web for information, please turn on the "Web Search" switch. Prompt example 1: Label each comment as praise, complaint, or question: First: Shipping was so fast! Second: The box arrived crushed. Third: Do you have it in black? Prompt example 2: Extract the name, phone number, and appointment date from this text into a table: Mr. Chen booked a house viewing on the afternoon of Nov 5, phone 0912-000-123. Prompt example 3: Summarize this notice in under 30 words: The library is closed Nov 10 to 12 for annual maintenance, and loans due during that time are extended by one week. Prompt example 4: Write a SQL query on the loans table that finds books more than 14 days overdue and not yet returned, showing reader name and book title, sorted by days overdue, longest first. Prompt example 5: Which FAQ best matches this customer's question? Reply with the number only. 1) Changing my password; 2) Shipping time; 3) Requesting an invoice. Question: When will yesterday's order arrive? Prompt example 6: Check which member records are invalid and explain why: Amy Lin / amy@example.com / 28; Ben Wu / ben@example / thirty; Cathy Chen / cathy@example.com / -5 Prompt example 7: Give each article a title of under 8 words: 1) A community library adds a family reading corner; 2) Night market vendors cut plastic; 3) High school students team up for a robotics contest. Prompt example 8: Split these addresses into city, district, road, and house number, one per line separated by commas: 100 Heping E. Rd. Sec. 2, Da'an Dist., Taipei; 99 Taiwan Blvd. Sec. 3, Xitun Dist., Taichung Prompt example 9: Does each pair ask the same thing? Answer only Same or Different. 1) How do I unsubscribe from the newsletter? / How can I stop the newsletter? 2) Is there store parking? / Are pets allowed? Prompt example 10: Sort this Python list by score from highest to lowest and reply with only one line of code: students = [('Amy', 82), ('Ben', 95), ('Cathy', 77)] Maximum context window (input and output combined): about 1,000,000 tokens. Maximum output per reply: about 64,000 tokens (model reasoning, if any, counts toward this limit). Maximum total upload size per request: 512MB (all files and message content combined). Accepted document formats: pdf, docx, xlsx, pptx, txt, csv (file content is provided to the model as extracted text). Accepted image formats: gif, jpeg, jpg, png, webp. Animated GIF images are analyzed using only the first frame. Maximum number of images: 600. When uploading more than 20 images, each side must be at most 2,000 pixels. Maximum size per image: 10MB (base64-encoded). Maximum total image size per message: 20MB (text and history share the same request size limit; uploads close to the limit may still be rejected). Image dimensions: 200 to 8,000 pixels per side. Output content formats: Text/Image Pricing: Low Continuous conversation: Supported AI can make mistakes. Please verify important information.