200K tokens
Context window
Anthropic Claude
Claude Opus
Long code review, editing, following long instruction lists
64K tokens
Max output
$15
Input per 1M
$75
Output per 1M
early 2025
Knowledge cutoff
text, image
Modalities
Last verified 20 May 2026 Report a correction
What it tends to refuse
Weapons and malware detail, sexual content involving minors, impersonating a named living person in persuasive text
Handles long inputs without losing the earlier instructions, which is the main reason people reach for it on code review. Costs enough that most teams route only the hard requests here.
Standing rules for this model go in CLAUDE.md, not in every message.
See the config files.
Written for Claude
All 13- CheatsheetUniversal
Forcing an output format that holds
The phrasings that actually constrain output, and the ones people use that do not. Tested by asking for the same thing twenty times and counting the…
Tested on Claude Sonnet · May 2026
Sysadmin who flags what cannot be undone
For infrastructure work: gives the command, but names anything irreversible before it, and never assumes a test environment.
Tested on Claude Sonnet · May 2026
Senior engineer who reads before changing
Turns the model into a colleague who answers the question you asked, touches only what you pointed at, and says out loud which parts it did not…
Tested on Claude Sonnet · May 2026
- Before and afterUniversal
Asking for a summary
The same document, the same model. The weak version returns the article rearranged; the strong one returns something you can act on, because it says…
Tested on Claude Sonnet · May 2026
- WorkflowUniversal
Find dead code across a repository
A four-step pass that maps the repo, lists exported symbols, checks each one for real callers and produces a deletion list you can actually act on…
Tested on Claude Sonnet · May 2026
- ConfigUniversal
Editor rules for a Django project
A rules file that pins the Django and Python versions, the app layout and the things an assistant should never regenerate.