Anthropic appears to be preparing the release of its Haiku 3.5 model, with recent updates in the codebase offering clues about its capabilities. Notably, references to tooltips suggest that Haiku 3.5 will lack image processing features at launch. This indicates that the initial rollout will likely focus on text tasks, without integrated vision capabilities.
New Claude experiments: "claude_ai_chili" and "claude_ai_cinnamon" - claude-3-5-haiku-20241022 is coming to claude[.]ai (no vision capabilities & no PDF uploads) - and "hidden" as "Claude 3 Haiku"? pic.twitter.com/ujVkjbFQwr
— Tibor Blaho (@btibor91) December 6, 2024
Speculation points to a potential release in December, as Anthropic may be aiming to respond to the wave of announcements from OpenAI. The timing aligns with the competitive landscape, where frequent advancements are shaping expectations for AI models.
In addition, work on the folder upload feature continues, with new tooltips clarifying that files uploaded through this feature won’t be used to train the model. Interestingly, the system creates a hidden project for every folder uploaded, hinting at a structured approach to managing and isolating user data. This development suggests Anthropic is prioritizing transparency and user control as it enhances Claude’s functionality.
These updates reflect Anthropic’s focus on refining features and meeting user needs while staying competitive in the evolving AI space.
