Anthropic Claude Haiku 5.5 is now available on Omnifact via Google Cloud and Anthropic.
Feature Drop: Claude Haiku 5.5 Now Available
Published on October 8th, 2026
Claude Haiku 5.5 is now available on Omnifact. Anthropic released it yesterday as its fastest and most capable small model yet, delivering a major performance jump at up to 90% lower token pricing than its predecessor. For the quick, high-volume work that fills most of a working day, such as summaries, translations, classifications, and first drafts, Haiku 5.5 is now our recommended default.
A Generational Jump for the Haiku Family
When we introduced Claude Haiku 4.5, it was the dependable choice for fast answers. Haiku 5.5 keeps that speed and closes much of the gap to the larger models. On Anthropic's published benchmarks, it more than doubles Haiku 4.5's score on GDPval-AA, a test of real-world knowledge work across professions, and lands within reach of Claude Sonnet 5.5.
The jump is largest in multi-step work. On OSWorld, which measures how well a model completes tasks on a computer, Haiku 5.5 reaches 72.4% compared to 15.7% for Haiku 4.5. That makes it a strong fit for agent tasks that run many quick steps in a row, like searching a Knowledge Base, reading the results, and condensing them into an answer.
Haiku 5.5 vs Haiku 4.5 at a Glance
| Area | Claude Haiku 5.5 | Claude Haiku 4.5 |
|---|---|---|
| Knowledge Work | 1620 on GDPval-AA v2.1 | 735 on GDPval-AA v2.1 |
| Computer Tasks | 72.4% on OSWorld 2.1 | 15.7% on OSWorld 2.1 |
| Token Pricing | $0.10 input / $0.50 output (per 1M tokens) | $1.00 input / $5.00 output (per 1M tokens) |
| Context and Output | 1M context / 128k output tokens | 200k context / 64k output tokens |
| Reasoning | Adjustable effort (Low to Extra High) | No adjustable effort |
| Knowledge Cutoff | June 2026 | July 2025 |
Token prices apply to prompts of up to 100,000 tokens. Longer prompts cost $0.50 input and $2.50 output per million tokens, which is still half the price of Haiku 4.5.
Built for Fast, Everyday Work
Haiku 5.5 suits high-frequency tasks where you need answers in seconds:
- Summaries and translations: Condense meeting notes, long email threads, or reports, and translate them for colleagues in other countries.
- Sorting and extraction: Classify incoming requests, pull key figures from documents, or turn unstructured text into clean lists.
- Large documents: The 1-million-token context window, five times that of Haiku 4.5, lets Haiku 5.5 work through long contracts, manuals, and document collections in one conversation.
Haiku 5.5 is also the first Haiku model with adjustable reasoning effort. Pick an effort level in the model picker in the Chat Input: keep it low for instant replies, or raise it when a question needs more careful thought. For complex analysis or demanding coding work, Claude Sonnet 5.5 and Claude Opus 5.5 remain the stronger choices.
Available Today with EU Data Residency
Claude Haiku 5.5 is available today for organizations on Omnifact via Google Cloud and Anthropic. Through Google Cloud, requests are processed within the EU. Every interaction remains protected under zero data retention commitments and Omnifact's Privacy Filter.
New organizations get Claude Haiku 5.5 by default as an included base-tier model in place of Haiku 4.5. Administrators of existing organizations can enable it in model settings to make it immediately available to everyone in their workspace. Switch to Claude Haiku 5.5 in the Chat Input model picker today to try it on your everyday workflows.
