Claude Opus 5.5 was published with a typical mission running cost reduction of 40%

September 23rd, Anthropic 22nd Claude Opus 5.5, this is Claude 5.5ModelI don't know. Officially, the new model reaches the level of Claude Fable 5.1 in most of its work, and the cost of processing typical loads under conventional settings is 40% lower than Opus 5 and output speeds above 30%。

Claude Opus 5.5 was published with a typical mission running cost reduction of 40%

Opus 5.5 Input, output and cache reading prices per 1 million Token are $4, $20 and $0.2 respectively. In the case disclosed by Anthropic, the early tester used it to complete a 680,000 line code migration within less than one day; in a web-based optimisation test, the model was successful 39 times in 40 attempts。

This is the first time that Anthropic has published a public call to slow the forward model. The model was evaluated by external institutions such as Frontier Design and METR, using biosafety and cybersecurity protection similar to Claude Fable 5.1. Claude Sonet 5.5 and Haiku 5.5 will be launched in the coming weeks。

statement:The content of the source of public various media platforms, if the inclusion of the content violates your rights and interests, please contact the mailbox, this site will be the first time to deal with.
Information

GPT-6 Sol and Luna published, API prices down 50%

2026-9-23 10:00:41

Information

Tenus: Siri AI should not replace relationships

2026-9-23 10:04:30

Search