Anthropic (private, ticker ANTH.PVT) unveiled its new high-performance large language model, Opus 5.5, on Tuesday. This launch marks the first model in the Claude 5.5 product line and comes shortly after CEO Dario Amodei publicly urged the industry to slow its pace of AI advancement.

According to Anthropic, the new model delivers overall performance on par with Fable 5.1 while cutting usage costs by 40% compared to its predecessor, Opus 5. Pricing has become a key competitive lever among AI companies. For Opus 5.5, the cost is set at 4 USD per million input tokens and 20 USD per million output tokens, down from 5 USD and 25 USD respectively for the earlier Opus 5. Tokens serve as the unit of measurement for language models, representing words or word fragments.

On the safety and alignment front, Anthropic states that the Claude 5.5 series has completed extensive alignment testing, external evaluations, and integrated safety guardrails for high-risk domains such as cybersecurity and biology. The company also notes plans to further refine rewards for AI alignment and develop novel safety training scenarios in the future.

The AI industry remains deeply embroiled in debates over existential risks tied to the technology. Earlier this month, Anthropic researcher Jacob Coxson announced his departure via a post on X. He argued that neither Anthropic, his current employer, nor OpenAI, where he previously worked, has genuinely prioritized AI safety. He contended that practitioners in the field truly believe the technology could wipe out humanity before the end of this decade. In response, Evan Hubinger, Anthropic's lead for AI alignment science, replied to Coxson's post, backing the former employee's stance and suggesting the probability of AI causing human extinction exceeds 10%.

Shortly after that exchange, Amodei published a lengthy statement urging the entire industry to slow down AI research and development. OpenAI CEO Sam Altman and SpaceX's Elon Musk both voiced support for Amodei's position. However, opposition has surfaced within the sector, with some critics arguing such risk concerns lack scientific grounding. President Trump has also publicly dismissed calls to decelerate AI progress.