Anthropic unveils Claude Opus 4.7
Anthropic announced Claude Opus 4.7, which the company and social discussion describe as its most powerful model to date and one being debated as a contender for leading model rankings. The rollout has prompted promotional integrations and chatter about its coding and reasoning capabilities on social channels. (x.com)
Anthropic said Thursday that Claude Opus 4.7 is now generally available, positioning it as the company’s strongest broadly released model. (anthropic.com) Anthropic said Opus 4.7 improves on Opus 4.6 in advanced software engineering, long-running coding tasks, instruction following, and image understanding at higher resolution. The company said developers can access it starting April 16, 2026 across Claude products, the Claude application programming interface, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. (anthropic.com) Pricing stayed at Opus 4.6 levels, with Anthropic listing $5 per million input tokens and $25 per million output tokens for the new model. Anthropic’s release notes also said the launch went live on April 16, 2026. (anthropic.com; support.claude.com) Large language models predict and assemble text, code, and images from patterns in training data, and vendors compete on how well those systems handle longer, messier tasks without drifting off course. Anthropic said Opus 4.7 is aimed at that race, with added emphasis on “complex, long-running tasks” and checks that let the model verify parts of its own work before answering. (anthropic.com) Anthropic tied the release to a separate safety effort around cybersecurity. The company said Opus 4.7 is “less broadly capable” than Claude Mythos Preview and is being used to test automated blocks on prohibited or high-risk cyber requests before any wider release of Mythos-class systems. (anthropic.com; cnbc.com) That places Opus 4.7 in the middle of Anthropic’s recent product arc. Anthropic introduced Claude 4, including Opus 4 and Sonnet 4, on May 22, 2025, then released Opus 4.6 in February 2026 before this April 16, 2026 update. (anthropic.com; cnbc.com) Amazon moved quickly to package the launch for cloud customers. Amazon Web Services said the model is available in Bedrock on April 16 and highlighted Anthropic-reported benchmark scores of 64.3% on SWE-bench Pro, 87.6% on SWE-bench Verified, 69.4% on Terminal-Bench 2.0, and 64.4% on Finance Agent v1.1. (aws.amazon.com) Anthropic also opened a Cyber Verification Program for security researchers who want to use Opus 4.7 for vulnerability research, penetration testing, and red-teaming. The company said lessons from that controlled use will inform whether it can eventually deploy more capable Mythos-class models more broadly. (anthropic.com) The immediate question is whether Opus 4.7’s coding and reasoning gains hold up outside Anthropic’s own tests. Anthropic released it as its top generally available model, but the company also said its more powerful Mythos system remains restricted. (anthropic.com; cnbc.com)
