Anthropic Launches Claude Opus 4.7: Better Code & Vision

Anthropic released Claude Opus 4.7 on April 16, 2026, delivering a meaningful upgrade to its flagship AI model with better coding, sharper vision and a new ability to double-check its own work. The release represents a notable improvement on Opus 4.6 in advanced software engineering, with particular gains on the most difficult tasks. The update arrives alongside major enhancements to Claude Code, including the introduction of “repeatable routines” and a redesigned workspace.

AI robot displaying major enhancements in IQ tests, advanced reasoning, and visual understanding capabilities for code ana...

Major Performance Improvements in Coding and Vision

The new model brings stronger software engineering capabilities, better vision, sharper instruction following, and more reliable long-running agent work. In benchmark testing, the update delivers a 13% lift on coding benchmarks and 3x more production tasks resolved.

Opus 4.7 accepts images up to 2,576 pixels on the long edge, roughly 3.75 megapixels, while Opus 4.6 topped out at 1.15 megapixels. This substantial vision upgrade enables screenshots, dense diagrams, design mockups, and documents to come through at actual fidelity.

According to Anthropic’s benchmarks, Opus 4.7 at full resolution scores 79.5% on visual navigation without tools versus 57.7% for Opus 4.6. The enhanced resolution particularly benefits computer use and document analysis workflows.

New xhigh Effort Level for Enhanced Reasoning

Opus 4.7 introduces a new xhigh (“extra high”) effort level between high and max, giving users finer control over the tradeoff between reasoning and latency on hard problems. Anthropic recommends starting with high or xhigh for coding and agentic use cases, and Claude Code now defaults to xhigh for all plans.

The performance gains are significant. At every effort level, Opus 4.7 outperforms Opus 4.6’s equivalent, with the new xhigh at 100k tokens scoring 71%, already ahead of Opus 4.6’s max at 200k tokens.

Task Budgets Beta and Enhanced Agentic Capabilities

Task Budgets is a brand-new feature in Claude Opus 4.7, allowing developers to set a consultative token budget for Claude to guide the model to allocate resources reasonably during long-running agentic tasks. Developers set a total token budget (minimum 20K tokens), the model sees a real-time countdown during execution, uses this to prioritize the most important work, and gracefully completes the task as the budget runs out.

Opus 4.7 handles complex, long-running tasks with rigor and consistency, pays precise attention to instructions, and devises ways to verify its own outputs before reporting back. The improvements in file system-based memory mean that agents that write to and read from scratchpads or notes files across long sessions get noticeably more reliable behavior, with multi-session work that previously lost context now holding it.

Claude Code Updates: Repeatable Routines and /ultrareview

Alongside the model release, Anthropic introduced significant updates to Claude Code. The platform now features repeatable routines, which enable AI-powered automation to streamline developer workflows. Claude Code also gains the new /ultrareview command that runs a dedicated review session that reads through your changes and flags what a careful reviewer would catch.

The workspace has been redesigned to improve efficiency, and auto mode in Claude Code is now available for Max plan subscribers, not just Teams/Enterprise/API customers.

Cybersecurity Safeguards and Mythos Comparison

Opus 4.7 is the first Claude model shipping with automated detection and blocking for prohibited cybersecurity uses, coming directly from last week’s Mythos Preview and Project Glasswing. The model is releasing with safeguards that automatically detect and block requests that indicate prohibited or high-risk cybersecurity uses.

Anthropic publicly conceded that the new Opus model does not match the performance of Mythos, showing in a chart that Opus 4.7 beats Opus 4.6, ChatGPT 5.4, and Google Gemini 3.1 Pro in a number of key benchmarks, but still falls short of its Mythos Preview model, which has only been released to a handpicked group of tech and cybersecurity companies.

Availability and Pricing

For developers interested in building AI solutions that demand frontier intelligence, Opus 4.7 is available on the Claude Platform natively, and in Amazon Bedrock, Google Cloud’s Vertex AI, and Microsoft Foundry. Pricing for Opus 4.7 starts at $5 per million input tokens and $25 per million output tokens, with up to 90% cost savings with prompt caching and 50% savings with batch processing.

For business users and consumers who want to collaborate with our most powerful model on complex tasks, Opus 4.7 is available on Claude for Pro, Max, Team, and Enterprise users.

Breaking Changes for Developers

The release includes important breaking changes that developers need to address. This is a major version upgrade that includes breaking changes including the removal of Extended Thinking Budgets, cancellation of sampling parameters, and a completely updated tokenizer.

Opus 4.7 uses an updated tokenizer that improves how the model processes text, with the tradeoff that the same input can map to more tokens, roughly 1.0 to 1.35× depending on the content type. Developers using temperature, top_p, top_k, or extended thinking budgets will need to migrate their code to avoid errors.

Key Facts

Released April 16, 2026 across Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry
13% improvement on coding benchmarks and 3x more production tasks resolved compared to Opus 4.6
Vision resolution increased to 3.75 megapixels (2,576 pixels long edge), up from 1.15 megapixels
New xhigh effort level provides finer control between reasoning depth and latency
Task Budgets beta feature enables resource allocation for long-running agentic tasks
Pricing remains at $5 per million input tokens and $25 per million output tokens
First Claude model with automated cybersecurity safeguards
Updated tokenizer increases token count by approximately 1.0 to 1.35×
Claude Code adds repeatable routines, /ultrareview command, and auto mode for Max subscribers