Opus 5.5 Took Fable's Job. Haiku 5.5 Proved Me Wrong (thoughts.jock.pl)

Anthropic has recently launched a trio of new AI models—Opus 5.5, Sonnet 5.5, and Haiku 5.5—just weeks apart, significantly enhancing their AI/ML offerings. Opus 5.5 has emerged as a preferred tool for various "agentic" tasks, demonstrating improved performance without the need for users to constantly adjust effort settings. Its pricing also dropped from $5 to $4 per million input tokens, making it more accessible. In contrast, the Fable 5.1 model has become less favorable due to its higher costs despite maintaining some strengths in design. Notably, Haiku 5.5 has exceeded expectations, showing impressive benchmarks and offering a highly competitive price of just $0.10 per million input tokens. It is designed for execution tasks, with Opus 5.5 handling planning and reasoning. The introduction of monthly API credits for subscribers adds significant value, allowing for more efficient use of Anthropic's models without incurring extra costs. Overall, these developments represent a substantial leap forward for Anthropic, positioning them strongly in the ongoing AI race.