Anthropic推出高性能低成本Claude Opus 5.5

According to MarktechPost, Anthropic has released the Claude Opus 5.5, the first model in the new Claude 5.5 family. The team claims it delivers Claude Fable 5.1-level performance on most tasks. At the same time, the operating costs of the model are 40% lower than those of the Opus 5 with standard settings. In Anthropic's own benchmarks, the model shows leadership in coding, computer use and knowledge work. Claude Opus 5.5 is available as an API-driven model, however model weights have not been released, eliminating the possibility of self-hosting. Developers can use claude-opus-5-5 on the Claude platform, Amazon Web Services, Google Cloud and Microsoft Azure. As with previous Opus models, zero data recording is available. Benchmarks have shown that the Opus 5.5 performs well, but doesn't always outperform other models. For example, at best effort the model uses adaptive thinking and production safety measures are enabled. On the FrontierCode v1.1 benchmark, Opus 5.5 scored 54.4%, which is higher than GPT-6 Astra's 53.3%, but at only a fifth of the cost per task. Regarding prices and speed, Opu

s 5.5 requires less computing power than Opus 5, and this is reflected in the prices. For 1 million tokens, the price is $4 per input and $20 per withdrawal, which is also 40% cheaper than Opus 5. Output generation is over 30% faster than Opus 5. Fast mode on the Claude platform offers speeds up to 2.5 times faster at $8 per input and $40 per withdrawal per million tokens. Early testers reported significant improvements. One tester completed the migration of 680,000 lines of code in less than a day, while a similar task with Opus 5 took more than 20 hours. Deloitte noted that Opus 5.5 at low effort detected 72% of known errors, while Opus 5 at high effort only detected 56%. Security and safety precautions have also been improved. Opus 5.5 performed best in Anthropic's automated behavioral audit, which covers nearly 2,000 scenarios. In the new containment test, the model tried to bypass restrictions 85% less often than Opus 5. Additionally, integration changes were made, such as the inability to turn off thinking and the presence of watermarks to comply with the EU AI Act. Thus, Claude Opus 5.5 is a demon

It offers high performance and significant cost reduction, making it attractive to developers and companies. The material was prepared by a neural network based on the original publication. AI may make mistakes in details (numbers, names, dates, quotes) - for important information, check with the original source. OpenAI introduces GPT-6 Sol and Luna with low price and fewer errors Anthropic introduced Claude Opus 5.5, beating Fable 5.1 in key indicators Kyutai released the Voice of Reason model for verbal solving of mathematical problems. Nokia opens access to AnyJev, a library for decision-making without training. OpenAI introduced GPT-6 Sol and Luna with new API pricing, reducing the cost by 50%.