Claude Opus 5.5 matches Fable 5.1 performance at lower cost and promises less "Claudish" writing

AI Digest

Anthropic推出Claude Opus 5.5，性能接近Fable 5.1但成本降低40%，运行更快，且写作更自然，同时强化安全防护。

Anthropic's Claude Opus 5.5 matches Fable 5.1 performance at 40% lower cost, with faster processing and reduced 'Claudish' writing, plus enhanced security measures.

Key points

• Opus 5.5性能接近Fable 5.1但成本降低40%，运行速度提升30% Opus 5.5 matches Fable 5.1 performance at 40% lower cost with 30% faster processing
• 价格下调20%，输入输出token成本分别降至4美元和20美元 Token pricing cut 20% to $4 input and $20 output per million tokens
• 在多个基准测试中超越GPT-6 Astra和GPT-5.6 Sol Outperforms GPT-6 Astra and GPT-5.6 Sol in multiple benchmarks
• 新增防数据蒸馏机制'Preserved Thinking'限制API编辑上下文 New 'Preserved Thinking' anti-distillation measure limits API context editing
• 强化生物安全与前沿LLM开发管控措施 Enhanced security controls for biology and frontier LLM development

Takeaway: Opus 5.5实现性能与成本的平衡，适合对效率敏感的商业应用场景。 / Opus 5.5 achieves performance-cost balance, ideal for efficiency-sensitive business applications.

Why it matters 展示AI模型在商业应用中的成本效率突破，对寻求性价比解决方案的开发者有实际参考价值。