Anthropic has released its new top-tier model, Claude Opus 4.5. The company says it sets records in software engineering benchmarks, runs more efficiently, and adds new control and agent features to the Claude platform.
Two months after shipping Sonnet 4.5, Anthropic is rolling out its next flagship: Claude Opus 4.5. Anthropic describes it as the world's most capable model for programming, autonomous agents, and computer control, with gains in routine tasks like spreadsheet editing, Deep Research, and slide creation.
Developers can access Opus 4.5 through the API, the Claude apps, and major cloud platforms. Pricing is 5 dollars per million input tokens and 25 dollars per million output tokens. The move is a response to growing price pressure in the market: Opus 4, released in May, was priced at 15 dollars per million input tokens and 75 dollars per million output tokens, so Opus 4.5 represents about a two-thirds price cut.
Anthropic is also adjusting usage limits. For people using Claude and Claude Code with access to Opus 4.5, the company is removing model-specific caps. Max and Team Premium users are receiving higher overall quotas so they effectively maintain the same usable token volume they had with Sonnet. These limits apply specifically to Opus 4.5 and will shift again as new models arrive.
