大模型
www.aibase.com ·
Claude Sonnet 5.5 发布：编程能力反超旗舰，API 价格腰斩
Anthropic 推出中端模型 Claude Sonnet 5.5，据称在多项代码生成与理解测试中超越旗舰 Opus 5.5，成为其产品线中编程能力最强的模型，通用能力也接近旗舰水平。API 输入输出价格分别降至每百万 token 2 美元和 10 美元，官方称整体使用成本可降低最多 30%。安全方面新增将高风险请求转交旧模型处理及防抄袭机制。
AI推荐理由 内容涉及 Anthropic 新模型 Claude Sonnet 5.5 的发布，包含编程能力、API 定价和安全策略等具体信息，对开发者和 AI 应用有参考价值。但来源为 AIbase 聚合站，非官方首发，正文未提供可验证的基准测试数据或官方链接，且发布时间为 2026 年，需谨慎对待。
62
正文
Claude Sonnet 5.5 Release: Programming Capabilities Surpass Flagship, API Price Halved
AIbase基地
Published inAI News · 3 min read · Sep 29, 2026
11
On September 29, AI company Anthropic officially launched its latest model, Claude Sonnet5.5, positioned in the mid-range market, yet it demonstrated capabilities approaching or even surpassing the flagship model Opus5.5 in various benchmark tests, drawing widespread industry attention.
According to official information, the biggest highlight of Sonnet5.5 is a significant improvement in programming capabilities. In multiple code generation and comprehension tests, the model outperformed its more expensive counterpart Opus5.5, becoming the model with the strongest coding ability in Anthropic's product line. At the same time, its general abilities such as knowledge answering and cross-domain reasoning have also basically caught up with the flagship level.
Anthropic Official Benchmark Test Comparison Chart
What has drawn even more market attention is the pricing strategy. The API call price of Sonnet5.5 has been halved—only $2 and $10 per million tokens for input and output respectively. Anthropic officials stated that due to the model's more efficient reasoning path, the overall usage cost can be reduced by up to 30%, significantly lowering the access barrier for developers and enterprises.
In terms of security, Anthropic has also upgraded its protection mechanisms. In response to Sonnet5.5's enhanced coding capabilities, the system has introduced a new strategy of automatically transferring high-risk requests to the older model for processing, and also added an anti-plagiarism mechanism to prevent model outputs from being maliciously reverse-engineered.
ClaudeSonnet5.5AIcompanyprogrammingabilitymodel
This article is from AIbase Daily
Scan to view
Welcome to the [AI Daily] column! This is your daily guide to exploring the world of artificial intelligence. Every day, we present you with hot topics in the AI field, focusing on developers, helping you understand technical trends, and learning about innovative AI product applications.
—— Created by the AIbase Daily Team
© Copyright AIbase Base 2024, Click to View Source -