Prime Big Deal DaysThe best deals, live through October 7Best deals, live now ›See the deals ›
October 8, 2026
Anthropic Launches Claude Haiku 5.5, Adds Monthly API Credits for Max and Team Users

Anthropic Launches Claude Haiku 5.5, Adds Monthly API Credits for Max and Team Users

Posted 41 minutes ago by

Anthropic has released Claude Haiku 5.5, its new small AI model for quick, high-volume tasks, with sharply lower API pricing. The launch also brings monthly API credits for Max and Team subscribers and a price cut for Sonnet 5.5 cache reads.

Anthropic's Claude Haiku 5.5 launch artwork

The model is aimed at jobs such as summarizing documents, classifying information, querying databases, and handling customer support. It can also take on smaller coding tasks within a larger AI workflow. For the first time in the Haiku lineup, developers can adjust effort levels to trade off cost against the model's reasoning capabilities.

Token pricing is 90% lower than Haiku 4.5 for prompts up to 100,000 tokens and 50% lower for longer prompts. That does not translate directly into the same savings on every task: the updated tokenizer uses slightly more tokens, and rates vary with prompt size. Anthropic estimates the model costs around 75% less to run on average.

The new API rates, per million tokens, are:

  • Input: $0.10 for prompts up to 100,000 tokens; $0.50 for longer prompts.
  • Output: $0.50 for prompts up to 100,000 tokens; $2.50 for longer prompts.
  • Cache Reads: $0.01 for prompts up to 100,000 tokens; $0.05 for longer prompts.
  • Cache Writes: $0.125 for prompts up to 100,000 tokens; $0.625 for longer prompts.

Separately, cache reads for Claude Sonnet 5.5 now cost $0.10 per million tokens, down from $0.20. The company estimates that this cut makes typical agent workloads about 20% cheaper to run.

Anthropic's benchmark results show gains over the previous small model. On the offline subset of OSWorld 2.1, which tests computer use, the score rose from 15.7% to 72.4%. Its GDPval-AA v2.1 score for knowledge work increased from 735 to 1620.

Anthropic's OSWorld 2.1 offline-subset computer-use benchmark compares accuracy and cost across effort settings
Anthropic's computer-use benchmark results across effort settings.

Haiku is still intended for narrower jobs. Sonnet and Opus 5.5 remain the better choices for complex coding work, according to Anthropic. The new model is available now through the Claude Platform, AWS, Google Cloud, and Microsoft Azure, with the API identifier claude-haiku-5-5.

Monthly API Credits for Subscribers

Starting this week, eligible subscribers can claim monthly API credits without a separate Console payment card. Max 5x users get $100 per month, while Max 20x users get $200. Team plans receive $20 per Standard seat and $100 per Premium seat, pooled across the organization and capped at $500 each month.

The credits can be used with any available model on the Claude Platform API. They do not increase the subscription's app usage limits or cover interactive Claude Code sessions, and they cannot be used through third-party cloud providers. Unused credits expire rather than rolling over. Eligible plans must have been active for at least seven days, and credits are claimed through Claude's website.

Add Comment
Would you like to be notified when someone replies or adds a new comment?
Yes (All Threads)
Yes (This Thread Only)
No
iClarified Icon
Notifications
Would you like to be notified when we post a new Apple news article or tutorial?
Yes
No
Comments
You must login or register to add a comment...
Recent. Read the latest Apple News.
RECENT
Tutorials. Help is here.
TUTORIALS
How To Force Restart iPhone [Video]
How To Turn Off iPhone [Video]
How To Back Up Your iPhone to an External Drive on Mac [Video]
Where To Download iPad Firmware Files From
Where To Download iPhone Firmware Files From
Deals. Save on Apple devices and accessories.
DEALS