Anthropic has quietly rolled out the Claude 3.5 Haiku AI model for users. On Thursday, numerous social media users began sharing updates about its presence in both the Claude web interface and mobile applications. The company claims that this latest iteration of the Haiku model is its fastest large language model to date. Additionally, in several performance benchmarks, the foundation model surpasses the capabilities of the Claude 3 Opus, which was the leading model in the previous generation. Importantly, Claude 3.5 Haiku is accessible to all Claude users, regardless of their subscription tier.
Anthropic Unveils Claude 3.5 Haiku
While no official announcement accompanied the launch of the new Haiku model, several users on X (formerly Twitter) noted its presence on both the web and mobile platforms. Staff members from Gadgets 360 independently confirmed that Claude 3.5 Haiku is now the default language model on the chatbot and is the only option available for users on the free tier.
Anthropic initially introduced the Claude 3.5 family of AI models in October, coinciding with the debut of the Claude 3.5 Sonnet. At that time, the company emphasized that the 3.5 Haiku model is its fastest offering. The latest upgrades feature reduced latency (enhanced response time), improved adherence to instructions, and precise tool utilization.
For businesses, the company noted that Claude 3.5 Haiku is particularly effective at user-facing products, specialized sub-agent tasks, and creating personalized experiences driven by large datasets.
In terms of performance, the new Haiku model achieved a score of 40.6 percent on the Software Engineering (SWE) benchmark, surpassing both the initial 3.5 Sonnet and OpenAI’s GPT-4o. It also outperformed GPT-4o mini in the HumanEval and Graduate-Level Google-Proof Q&A (GPQA) benchmarks.
Earlier this month, Anthropic optimized the Claude 3.5 Haiku for the AWS Trainium2 AI chipset and incorporated support for latency-optimized inference in Amazon Bedrock. The company has yet to provide support for Google Cloud’s Vertex AI. The new AI model can generate text but is equipped to accept both text and images as input.
