release
Apr 16, 2026
By Teun
Claude Opus 4.7 Released — Anthropic's Most Capable Model Yet
Anthropic shipped Opus 4.7 with stronger coding, better vision, and improved instruction-following. Same pricing as Opus 4.6. Claude Code got Auto mode and a new xhigh effort level.
Anthropic has released Claude Opus 4.7, its latest flagship model, with stronger coding performance, better vision, and improved instruction following. The company says the new version is its most capable model yet, and it keeps the same price as Opus 4.6, at $5 per million input tokens and $25 per million output tokens.
For teams using Claude as part of software work, the most relevant change is the jump in coding quality. Anthropic says Opus 4.7 is better at complex software engineering tasks and long-running agent workflows, which matters when a model has to keep track of steps, tools, and context over time rather than just answer a single prompt.
⚡ New to this?
Anthropic has updated Claude Opus, which is one of its top AI models, with better coding, vision, and instruction-following. For a non-expert, that means the model should be better at tasks like writing software, reading screenshots, and sticking to exact directions. Claude Code, Anthropic's coding tool, also got new modes that change how much effort it spends on a task.
🦞 OpenClaw angle
If you run self-hosted agents through Anthropic, update your model routing to test Opus 4.7 on the workflows where mistakes are most expensive, especially multi-step coding and tool-using jobs. Keep the old model available behind a fallback flag until you have side-by-side logs, then compare task completion, tool-call accuracy, and format adherence on the same prompts. If you use visual inputs, add a small regression set of screenshots or diagrams so you can verify whether the vision upgrade actually helps your automation.
Instruction following also got better, according to Anthropic. That usually means the model is less likely to drift from constraints, miss a format requirement, or improvise when a workflow expects a specific sequence of actions. In agent systems, that difference often matters more than raw benchmark performance because small misses can cascade into broken automation.
Vision is another area where Anthropic says Opus 4.7 improved. For users building multimodal systems, that can matter when the model needs to read screenshots, inspect diagrams, or interpret UI states as part of an automation loop. The release note points to a broader push toward models that can handle both text and visual inputs more reliably in production workflows.
Anthropic also updated Claude Code alongside the model launch. The company says Auto mode is now available for Max subscribers using Opus 4.7, and there is a new xhigh effort level between high and max. Anthropic also added a /tui fullscreen command, which makes the terminal interface easier to use as a full-screen coding environment.
The pricing decision is notable because it removes a common barrier to upgrading. Better performance usually comes with a higher bill, but Anthropic kept the same Opus 4.6 price while shipping the new model, which makes the switch simpler for teams already budgeting around Anthropic usage. The docs page for Claude models now lists Opus 4.7 as the current top-end option, so the next step for users is to compare its behavior against their existing production prompts and agent flows.
