Claude 3.5 Haiku is now generally available for Claude users
Anthropic has confirmed the new version of its smallest model, Claude 3.5 Haiku, is now available in the Claude web experience and mobile apps. First announced in October, Claude 3.5 Haiku is intended to deliver Claude 3 Opus' performance at Claude 3 Haiku's speed.
Anthropic unveiled Claude 3.5 Haiku in October alongside an updated Claude 3.5 Sonnet. In that announcement, Anthropic highlighted that the new Claude 3.5 Haiku would deliver the same benchmark performance as Claude 3 Opus, the largest model from the previous version of the Claude family, with the speed of Claude 3 Haiku. Additionally, Claude 3.5 Haiku was endowed with a larger output window and extended knowledge cutoff (both compared to Claude 3 Haiku). Anthropic recommends Claude 3.5 Haiku for tasks including coding recommendations, data extraction and labeling, and content moderation.
In the original announcement, Anthropic stated it planned to make Claude 3.5 Haiku on its API, Amazon Bedrock, and Google Cloud’s Vertex AI by the end of October. However, the company did not mention whether the model would be available in the Claude AI assistant experience. On Thursday, reports on social media platforms started noting that the model was finally available on the Claude apps and web experience. Anthropic has since confirmed the general availability of Claude 3.5 Haiku for all Claude users.
Consistent with the October announcement, Claude 3.5 Haiku is text-only, so it doesn't support images as input—trying to attach a picture while using 3.5 Haiku, Claude automatically fell back to Claude 3 Haiku to process an attached file. The company stated it would enable Claude 3.5 Haiku to admit images as inputs, but provided no set timeline for this.
