Anthropic Releases Claude Opus 4.7
Anthropic released Claude Opus 4.7 on April 16, 2026. The model is available across all Claude products, the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.
What Is New
Opus 4.7 improves on Opus 4.6 in software engineering, with the biggest gains on harder tasks. It is the first Claude model with high-resolution image support. The maximum image size is now 2576 pixels or 3.75 megapixels, so the model can read small text, charts, and diagrams more reliably.
The model keeps the 1M token context window, 128k max output tokens, adaptive thinking, and the same tool and platform features as Opus 4.6.
Safety Changes
Opus 4.7 adds automatic safeguards that detect and block prompts linked to prohibited or high-risk cybersecurity work. This is a direct response to the concerns Anthropic flagged around its more capable Mythos Preview model, which remains restricted to a small set of partners under Project Glasswing.
How It Compares to Mythos
Anthropic said Opus 4.7 is not as broadly capable as Mythos, but it is safer to ship to the public. Mythos can find and exploit software vulnerabilities on its own in controlled tests. Opus 4.7 is tuned so that kind of behavior is blocked before it runs.
Why It Matters
The launch shows how Anthropic is splitting its roadmap. Public models get the general upgrades. The most powerful research models stay behind a gate because of cyber risk. For developers and teams, Opus 4.7 is the new default for complex reasoning, long agent runs, and visual tasks on the Claude Platform.
Discussion
Sign in to comment. Your account must be at least 1 day old.