Anthropic has announced the release of Claude Haiku 4.5, which provides similar coding performance to Claude Sonnet 4, but is twice as fast and one-third as expensive.
This follows the release of Claude Sonnet 4.5 two weeks ago, which Anthropic claimed as “the best coding model in the world” when it was announced. Claude Haiku 4.5 outperforms Sonnet 4 on SWE-bench, but falls second behind Sonnet 4.5.
“Claude Sonnet 4.5, released two weeks ago, remains our frontier model and the best coding model in the world. Claude Haiku 4.5 gives users a new option for when they want near-frontier performance with much greater cost-efficiency,” Anthropic wrote in a post.
The company explained that these two models can be used together to get the best of each, such as using Sonnet 4.5 to break down a complex problem into multiple steps and then having multiple Haiku 4.5 instances working together to execute those subtasks in parallel.
According to Anthropic, Haiku 4.5 excels at real-time, low-latency tasks, such as chat assistants, customer service agents, and pair programming.
It will cost $1 per million input tokens and $5 per million output tokens. For comparison, Claude Sonnet 4 and 4.5 cost $3 per million input tokens and $15 per million output tokens.
Claude Haiku 4.5 is available now in Claude Code and other Anthropic apps, as well as through the API, Amazon Bedrock, and Google Cloud’s Vertex AI.
