Claude Opus 5 ships with a 1-million-token context window as both default and maximum, 128,000 output tokens in the standard API, and unchanged Opus pricing of $5 per million input tokens and $25 per million output. [1]
Thursday said the context window enters daily use. Friday is still use. Opus 5 replaces Opus 4.8 as the workhorse. Fable 5 remains the frontier flagship. Mythos 5 stays ahead on cybersecurity. [1]
Anthropic launched the model on July 24. It is the new default on Claude Max and the strongest model on Claude Pro. Fast mode runs about 2.5 times quicker at twice the price. [1]
The company leans on Frontier-Bench, CursorBench, ARC-AGI 3, GDPval-AA, OSWorld. It says Opus 5 more than doubles Opus 4.8 on Frontier-Bench at a lower cost per task. Independent testers at Artificial Analysis ranked it first on intelligence and agentic indexes at launch. [1][2]
A million tokens is enough, in the marketing sentence, for a large codebase or a legal discovery set in one prompt. The production question is whether instruction-following holds at the far end of that window. Customers will find out on invoices. X asks who can pay because a million tokens at $5 in and $25 out is not a toy bill. [1]
Opus 4.1 is being retired. Automatic fallbacks send safety-flagged Opus 5 calls to another model instead of a hard block. Mid-conversation tool changes will no longer bust the prompt cache. That is the production tier talking. [1][3]
Labs sell a window. Teams that paste an entire repository into a prompt will learn whether the model still knows which file matters. That is the only test that counts. A window is a capability. A price is a policy. Opus 5 is both, and still not the top of the catalog.
Early-access notes are the usual parade. Cognition said Opus 5 approached Fable-level work on FrontierCode at half the cost. Cursor's Sualeh Asif said it sat just under Fable 5 on CursorBench. Zapier said it completed a churn-prevention sequence no prior model passed. Those are vendor-selected stories. They are still more useful than a context-window meme. [1]
MarkTechPost filed the launch as frontier-class coding at unchanged Opus rates. That hed is closer to Anthropic's own hierarchy than the 1M-token posts. Friday's daily-use question is whether a legal team or a repo owner will keep the whole pile in one prompt after the first invoice. X is right to ask who can pay. The lab is right that the workhorse, not the flagship, is what most desks will open. [3]
Effort settings let a customer trade tokens for depth. On some high settings, thinking cannot be switched off. That is how a million-token window becomes a daily habit rather than a stunt: you leave the pile in the prompt and pay for the thinking you asked for. Or you do not, and the meme stays a meme. [1][2]
Friday still has no public invoice that proves a legal team will keep a million tokens in one prompt. The workhorse is what desks will open. The crown is what X will quote. Those are different bills.
-- KENJI NAKAMURA, Tokyo