On October 22, 2024, Anthropic announced the launch of an improved version of its AI model, Claude 3.5 Sonnet, along with a new model named Claude 3.5 Haiku. The enhanced Sonnet model offers better coding capabilities, achieving a score of 49% on the SWEbench Verified benchmark, significantly up from 33.4%.
Claude 3.5 Sonnet operates twice as fast as its predecessor, Claude 3 Opus, and is designed for complex tasks such as contextual customer support and multi-step workflow orchestration. The new Claude 3.5 Haiku model matches the performance of Claude 3 Opus while being cheaper and faster.
A key feature introduced is the “computer usage” capability, allowing the AI to interact with computers similarly to humans. This includes tasks like web browsing, form filling, and data manipulation. Initial users include Asana, DoorDash, and Replit. However, Anthropic notes this feature is still experimental and may have errors.
The Claude 3.5 Haiku model is set to launch later this month, initially supporting text input only, with image input to follow. Developers can access the computer usage feature via Anthropic’s API, Amazon Bedrock, and Google Cloud's Vertex AI.
Overall, the updates signify a substantial advancement in AI capabilities, particularly in coding and task automation, with strong initial feedback from users indicating significant improvements over previous models.