Changes in the system prompt between Claude Opus 4.6 and 4.7
Source published ·Modelwire updated
Original coverage: Simon Willison ↗·How Modelwire adds context
The development
Anthropic released Claude Opus 4.7 on April 16, 2026, with updated system prompts compared to the February 4.6 version. The company continues its practice of publicly archiving system prompt changes, enabling transparency into how model instructions evolve across releases.
Modelwire’s AI-generated summary of coverage from Simon Willison.
Modelwire analysis
ExplainerOur AI-generated reading of the wider context and the next developments to watch.
The real story isn't the version bump itself but the practice: Anthropic publicly archives system prompt changes between releases, which means researchers and developers can track exactly how Anthropic's behavioral guardrails and default instructions shift over time, not just infer it from model outputs.
This lands in the middle of a dense week for Anthropic. The cybersecurity model Claude Mythos Preview (covered here from The Verge, April 17) and Claude Design (TechCrunch, April 17) both represent new product surfaces, each of which would carry its own system prompt logic. Watching how those prompts evolve across releases is one of the few concrete ways outsiders can audit whether Anthropic's stated safety priorities are actually reflected in deployed instructions, or whether they drift quietly under commercial pressure. That context makes the transparency practice more consequential than a routine changelog.
If the Opus 4.7 system prompt changes show expanded permissions around agentic or desktop-control behaviors (relevant given the Claude Code competition with OpenAI's upgraded Codex), that would signal Anthropic is loosening constraints to stay competitive. A diff showing tighter restrictions would suggest the opposite pressure is winning.
This interpretation is generated from the summary above and available source metadata. Our methodology · Report an error
MentionsAnthropic · Claude Opus 4.7 · Claude Opus 4.6 · Claude.ai · Simon Willison
How this coverage is produced
Modelwire uses AI to generate summaries and context from source headlines, snippets, and selected archive coverage. Automated checks do not verify every claim, and items are not routinely reviewed by a person before publication. Zacaria Solis operates the site. Read the linked source for the full evidence and report errors through our corrections process.
Modelwire summarizes, we don’t republish. The full content lives on simonwillison.net. If you’re a publisher and want a different summarization policy for your work, see our takedown page.