Overview
- Anthropic has introduced Claude Sonnet 5.5, priced at $2 per million input tokens and $10 per million output tokens, which is notably cheaper than Opus 5.5.
- This new model scored 70.6% on the Terminal-Bench 4.0 test, outperforming Opus 5.5, which achieved 66.4%.
- While independent evaluation by Artificial Analysis gave Sonnet 5.5 a score of 63.6%, it noted the model's higher token usage compared to others tested.
The release of Claude Sonnet 5.5 marks an upgrade from its predecessor, presenting a performance improvement of over 30%. According to Anthropic, the model excels in tasks like debugging and document creation, and shows significant design capabilities.
Sonnet 5.5 is particularly strong in coding tasks, as evidenced by its performance on Terminal-Bench 4.0. The model managed to complete a greater number of tasks efficiently, making it a more economical choice compared to its predecessors and competitors.
Independent assessments reinforced Sonnet 5.5's capabilities, achieving notable scores against Opus 5.5 and OpenAI's GPT-6 Astra. Despite claims of cost savings, the model's output at maximum effort incurs higher expenses, potentially counteracting its affordability at full capacity.
