Anthropic releases Claude Opus 5, a near-Fable-5 model at half the price

Anthropic shipped Claude Opus 5 on July 24, 2026 — the new head of the Opus tier and, as of today, the default model on Claude Max. List price is $5 per million input tokens and $25 per million output — half the Fable 5 rate — with a 1M-token context window and 128K output. Anthropic's framing is capability-per-dollar: near-flagship intelligence one rung below the Mythos-class Fable 5, with an effort dial that trades cost against capability per request.

The price is the story with history behind it. The Opus tier launched at $15/$75 with Claude 3 Opus in March 2024 and held there for twenty months; Opus 4.5 reset it to $5/$25 in November 2025, and it has not moved since — through 4.6, 4.7, 4.8, and now Opus 5. The strongest model below Anthropic's flagship costs a third of what the same slot cost a year ago, while the slot's capability has climbed within reach of Fable 5 itself.

For builders, the release changes API behavior, not just weights. Extended thinking is now on by default, a new max effort setting tops the reasoning dial, and there is a documented breaking change around the xhigh and max effort settings that returns HTTP 400 where earlier models accepted the request — worth reading Anthropic's what's-new page before switching model strings. Also in the notes: a 512-token prompt-cache minimum, an optional fast mode, Opus 4.8 staying on as the fallback, Opus 4.1 retiring August 5, and a deprecation-policy floor that pins Opus 5's earliest retirement at July 24, 2027.

The wider frame: Opus 5 is the third frontier-tier launch in fifteen days, after GPT-5.6 went GA on July 9 and Gemini 3.6 Flash shipped on July 21. Anthropic also leads with safety claims — an alignment audit and roughly 85% fewer cyber-classifier interventions than Opus 4.8 — landing while the company is reportedly scheduling investor meetings toward a fall Nasdaq listing.