Anthropic Launches Opus 5.5 at Reduced Prices with Fable-Level Performance

Anthropic has announced that its latest generative AI model—named Claude 3.5—has outperformed all previous versions in the company’s internal benchmark suite. In a statement released this week, the company said the new model is “the strongest‑performing model we’ve tested to date,” citing significant gains in language understanding, reasoning, and safety alignment.

Claude 3.5 builds on the same transformer architecture as its predecessor but incorporates a larger training corpus and a refined reinforcement‑learning‑from‑human‑feedback (RLHF) pipeline. In standardized tests such as the LAMBADA, MMLU, and a proprietary safety evaluation, the model achieved scores that exceed the prior best by 12 % on average. Anthropic also reported that the model’s response latency remains comparable to earlier releases, and that its safety filters have been tightened to reduce hallucinations and biased outputs.

The company’s announcement comes as the broader AI industry pushes for higher‑quality, more reliable large language models. By positioning Claude 3.5 as its top performer, Anthropic signals its intent to compete more directly with leading vendors while emphasizing its commitment to safety and alignment. The firm plans to roll out the model to enterprise customers later this year and to continue refining its architecture based on user feedback.