What's new in Claude Sonnet 5

The more consequential detail buried in the framing is that Anthropic is effectively deprecating the performance gap between its own tiers, which compresses the internal justification for Opus-level pricing and forces the question of what Opus 5 will need to deliver to hold its price point.

Modelwire has no prior coverage to anchor this to directly, so this story sits largely on its own in our archive. That said, it belongs to a broader pattern visible across the frontier lab space: the commoditization pressure on mid-tier models has been building as inference costs fall and competition on price-per-token intensifies. Anthropic's move here is less about capability announcement and more about defending developer mindshare against OpenAI's GPT-4.1 and Google's Gemini Flash lines, both of which have been pushing hard on the efficiency-first framing. The compression of the Sonnet-to-Opus performance gap is a structural change to Anthropic's own product ladder, not just a pricing adjustment.

Watch whether Anthropic raises Opus 5 capability claims significantly at its next release to re-establish tier separation. If Opus 5 ships within six months without a clear benchmark gap over Sonnet 5, that confirms the tier distinction is collapsing into a pure pricing fiction.