Claude Opus 4.7 Is Here: Anthropic's Most Capable Coding Model Yet
Anthropic has released Claude Opus 4.7, delivering major gains in software engineering, vision, and long-horizon agentic tasks — plus new cybersecurity safeguards learned from the Mythos Preview rollout.
In this article
What's New in Opus 4.7
The headline improvement is in software engineering. Opus 4.7 handles complex, long-running coding tasks with a rigor and consistency that previous versions couldn't match. According to Anthropic, users are now able to hand off their hardest coding work — the kind that previously needed close supervision — with confidence.
Three capabilities define the upgrade:
- Self-verification: The model devises ways to check its own outputs before reporting back, dramatically reducing silent errors.
- Instruction precision: It pays closer attention to nuanced directives and follows them more faithfully across long contexts.
- Sustained reasoning: Performance doesn't degrade over extended multi-step runs — a critical improvement for agentic workflows. (See also: Claude Code Routines) (See also: Claude Design) (See also: Project Glasswing) Project Glasswing: How Anthropic Is Using Claud...
Better Vision, More Creative Output
Beyond coding, Opus 4.7 brings substantially better vision: it processes images at higher resolution, which matters for tasks like reading chemical structures, interpreting complex technical diagrams, and understanding dense charts.
The model is also described as more tasteful and creative on professional tasks — producing higher-quality interfaces, slide decks, and documents. This is a subtler improvement than raw benchmark gains, but one that practitioners who use Claude for output generation will notice immediately.
Benchmark Numbers from Early Access Partners
Anthropic shared a range of benchmark results from enterprise partners who had early access. The numbers paint a consistent picture of meaningful gains:
- Cursor — On CursorBench, Opus 4.7 clears 70% vs Opus 4.6 at 58%.
- Rakuten — Resolves 3x more production tasks than Opus 4.6, with double-digit gains in Code Quality and Test Quality.
- Notion — +14% over Opus 4.6 at fewer tokens and a third of the tool errors. First model to pass their implicit-need tests.
- Harvey — 90.9% substantive accuracy on BigLaw Bench at high effort, with better handling of ambiguous document editing.
- Devin (Cognition) — Works coherently for hours, pushes through hard problems, unlocking deep investigation work not previously reliable.
Multiple partners independently noted that Opus 4.7 catches its own logical faults during the planning phase — a qualitative shift that's hard to capture in benchmarks but deeply meaningful for production use.
Cybersecurity Safeguards: A New Framework
Opus 4.7 is the first model to carry Anthropic's new cybersecurity safeguard framework, which automatically detects and blocks requests indicating prohibited or high-risk cyber uses. This is a direct consequence of what Anthropic learned from the limited Mythos Preview rollout and their Project Glasswing announcement.
Importantly, Opus 4.7's cyber capabilities are intentionally less advanced than Mythos Preview — Anthropic experimented with differential capability reduction during training specifically around cybersecurity. The goal is to build, test, and validate safeguards on this less-powerful model before eventually applying them to Mythos-class releases.
For security professionals doing legitimate work (vulnerability research, penetration testing, red-teaming), Anthropic has launched a new Cyber Verification Program at claude.com/form/cyber-use-case.
Availability and Pricing
Opus 4.7 is available today across:
- All Claude products (claude.ai)
- Anthropic API — model ID:
claude-opus-4-7 - Amazon Bedrock
- Google Cloud Vertex AI
- Microsoft Foundry
Pricing is unchanged from Opus 4.6: $5 per million input tokens and $25 per million output tokens.
What This Means for Claude Users
For most Claude users, the most relevant improvements are practical:
If you're a developer, the self-verification and sustained reasoning gains mean you can push harder tasks to the model and trust that it will flag its own uncertainty rather than confidently producing wrong output. The Cursor and Notion numbers suggest real workflow acceleration for code review and complex agentic pipelines.
If you use Claude for document and creative work, the better vision and more tasteful creative output are immediately useful. Higher image resolution support means better handling of diagrams, charts, and screenshots you paste in for context.
If you're building agents, the long-horizon consistency improvement is the most important feature. Devin and Notion both noted that Opus 4.7 is the first model to reliably sustain quality over multi-hour, multi-step agentic runs — a meaningful unlock for autonomous workflow design.
The cybersecurity safeguard framework is largely invisible to most users but represents Anthropic's responsible path toward eventually releasing Mythos-class capabilities more broadly. Watch for how these safeguards evolve — they're the foundation for the broader Mythos release when it comes.
Get the Claude playbook in your inbox.
One weekly email for Claude and Claude Code users. Real workflows, no hype. Subscribe and we send you The Claude Power-User Cheatsheet.
— ¶ —
Luke Thompson
Luke Thompson is the founder of The Operations Guide, LLC and editor of The Claude Insider. Based in Jonesborough, Tennessee, he has spent years building AI-augmented business systems and automation workflows for operators and teams. He began working with large language models in production well before the current wave of consumer AI tools, integrating them into client workflows, content pipelines, and operational infrastructure. At The Claude Insider, he writes about Claude with the specificity of someone who uses it daily as a professional tool — not as a reviewer or commentator, but as a builder. His coverage focuses on what actually works: prompt patterns, API integration strategies, agentic workflows, and the real-world tradeoffs that practitioners face. He is not affiliated with Anthropic, PBC.
Articles are researched and drafted with AI assistance, reviewed and edited by Luke Thompson.
Know where AI can pay off in your company.
Take the free two-minute AI Readiness Assessment. See your score, the two gaps holding you back, and the next move worth making.
