Anthropic Launches Claude Opus 5.5 With Stronger Coding Performance and Lower Pricing
aiinsider ·
Anthropic has officially launched Claude Opus 5.5, a new model tier that sets a state-of-the-art benchmark for coding and knowledge work tasks. This release positions Opus as the most capable option in the Claude lineup, surpassing its larger Fable model on various benchmarks and completing specific informal tasks that previous versions could not handle. The company highlights significant improvements in efficiency, noting that output tokens are now priced at $20 per million, down from $25, which reflects reduced compute requirements. Additionally, the model is faster and utilizes clearer language by leading with key information rather than excessive jargon. Due to its advanced capabilities in biology and cybersecurity, which match those of Mythos, Opus 5.5 is subject to the same safety safeguards as Fable, restricting uses like exploit development or biological weapon creation. This launch follows CEO Dario Amodei’s recent endorsement of pacing AI frontier progress to ensure risk prevention keeps pace with capability growth. Independent organizations such as METR and Frontier Design evaluated the model prior to its public release. Anthropic emphasized the need for greater public policy involvement in AI safety and promised to share more details about its safety efforts soon.
aiinsider