Anthropic releases Claude Opus 4.5 at a lower price than earlier Opus models

Anthropic/Claude models

Anthropic released Claude Opus 4.5 at $5 per million input tokens and $25 per million output tokens, positioning it as its best model for coding, agents, and computer use. It uses 48–76% fewer tokens than predecessors while matching or exceeding prior performance across vision, reasoning, and coding.

• State-of-the-art on SWE-bench Verified for real-world software engineering; scores higher than humans on Anthropic's internal 2-hour coding exam.
• Most robustly aligned model released to date, with superior resistance to prompt injection than other frontier models.
• New effort-control parameter balances capability versus efficiency; extended conversation support in Claude apps.