Claude Opus 4.7 Is Now Available

Anthropic released Claude Opus 4.7 on April 16, 2026. It's a direct upgrade to Opus 4.6, with meaningful gains in software engineering, vision quality, instruction following, and memory across long-running tasks. If you use Claude for serious production work, this one's worth your attention.

Here's what changed.

Claude Opus 4.7 Software Engineering Improvements

The headline upgrade is coding. Opus 4.7 is significantly better at complex, multi-step software engineering tasks — specifically the kind that previously required close human supervision. Early testers report being able to hand off hard coding work with more confidence and less babysitting.

The model is more rigorous on long-running tasks, follows instructions more precisely, and now verifies its own outputs before reporting back. That last part matters. Self-verification at the model level is a meaningful step toward actually autonomous engineering work.

One caveat: because instruction-following is now more literal, prompts written for Opus 4.6 may behave differently. If something looks off after migration, re-tune the prompt — don't assume it's a model failure.

Upgraded Vision: High-Resolution Image Processing

Opus 4.7 can now accept images up to 2,576 pixels on the long edge — roughly 3.75 megapixels, more than three times the resolution of prior Claude models. This opens real use cases: computer-use agents reading dense screenshots, data extraction from complex charts and diagrams, design review requiring pixel-level reference. For any workflow that depends on fine visual detail, this is a substantive improvement.

Note that higher-resolution images consume more tokens. If your use case doesn't require the extra fidelity, downsample before sending.

Smarter Agentic Memory and Long-Session Reliability

Opus 4.7 is better at using file system-based memory. It retains important notes across multi-session work and applies them to new tasks without requiring heavy re-contextualization at the start of each run. For teams running extended agentic workflows — think multi-day coding projects, ongoing research, or complex document processing — this materially reduces the overhead of keeping the model oriented.

New Effort Levels and Developer Controls in Claude API

Anthropic introduced a new xhigh effort level, slotting between high and max. This gives developers finer control over the reasoning-vs-latency tradeoff on hard problems. In Claude Code specifically, the default effort has been raised to xhigh for all plans.

Also shipping: task budgets in public beta on the Claude API, giving developers a way to guide token spend across longer runs. For teams managing cost at scale, this is a practical addition.

Claude Code Gets /ultrareview and Auto Mode

Two notable Claude Code updates alongside the model launch:

The /ultrareview slash command runs a dedicated review session that reads through code changes and surfaces bugs and design issues the way a careful human reviewer would. Pro and Max users get three free ultrareviews to try it.

Auto mode — which lets Claude make permission decisions autonomously for longer tasks with fewer interruptions — is now available to Max users. Used thoughtfully, it reduces friction on complex workflows without requiring you to grant blanket permissions upfront.

Cybersecurity Safeguards and the New Cyber Verification Program

Anthropic is using Opus 4.7 as a test bed for new cybersecurity safeguards before any broader release of Mythos-class models. Opus 4.7's cyber capabilities are intentionally constrained relative to Claude Mythos Preview, and the model ships with automatic detection and blocking of prohibited or high-risk cybersecurity requests.

Security professionals with legitimate use cases — vulnerability research, penetration testing, red-teaming — can apply through Anthropic's new Cyber Verification Program for appropriate access.

Opus 4.7 Pricing and Availability

Opus 4.7 is available now across all Claude products, the API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. Pricing stays the same as Opus 4.6: $5 per million input tokens, $25 per million output tokens. API access uses the model string claude-opus-4-7.

One migration note: Opus 4.7 uses an updated tokenizer that can map the same input to 1.0–1.35× as many tokens, depending on the content type. When output increases at higher effort levels, token usage can shift. Anthropic recommends measuring against real traffic and using the effort parameter or task budgets to tune accordingly.

What Claude Opus 4.7 Means for Marketing and Growth Teams

For marketing and growth teams using Claude in production, the practical implications are real. Better instruction-following means more predictable outputs from established prompts — once you've tuned them for the new literalness. Improved vision support opens up use cases in creative review, asset analysis, and multimodal content workflows. And enhanced agentic memory means extended tasks — think content pipelines, research sprints, competitive analysis — require less hand-holding across sessions.

AI is moving fast, and knowing which upgrades actually matter to your workflow versus which are noise is increasingly part of the job. If you're building AI into your content and growth strategy and want to make sure you're using the right tools the right way, our team at Winsome Marketing works with growth leaders on exactly that. Let's talk.