Anthropic releases Claude Opus 4.7 AI model

Anthropic released its Claude Opus 4.7 artificial intelligence model on Thursday, expanding public access to the latest version of its Claude Opus line. The company said the model is now available broadly through Claude AI, the Claude API, and via partners including Microsoft Foundry.

The company positioned Claude Opus 4.7 as an update aimed at multi-step reasoning and advanced coding work. Anthropic said the model improves performance in advanced coding tasks, visual intelligence, and document analysis, and described it as more effective for professional use cases such as producing higher-quality interfaces, slides, and documents.

Anthropic said Claude Opus 4.7 is priced the same as its predecessor, Claude Opus 4.6. At the same time, the company flagged a trade-off tied to how the model operates: Opus 4.7 may require more output tokens because it applies higher effort levels when processing tasks. That detail matters for users and organizations that manage usage through token-based limits or budgets, particularly when deploying the model through the Claude API or partner platforms.

Benchmark results cited by Anthropic showed Claude Opus 4.7 posting a 46.9 percent score on Humanity's Last Exam without tools. In the same comparison, Gemini 3.1 Pro scored 44.4 percent and GPT-5-4 Pro scored 42.7 percent, while Anthropic’s unreleased Claude Mythos scored 56.8 percent.

In testing with tools enabled, Anthropic reported GPT-5-4 Pro at 58.7 percent and Claude Opus 4.7 at 54.7 percent. Both trailed Claude Mythos, which was listed at 64.7 percent in the same tool-enabled setting.

The release adds another option for developers and enterprises choosing among leading AI systems for coding, document workflows, and multimodal tasks such as visual intelligence. Availability through Claude AI, the Claude API, and Microsoft Foundry also highlights how distribution partnerships can shape adoption across regions and industries, especially where organizations standardize on specific cloud and developer ecosystems.

Several uncertainties remain based on what was disclosed. Anthropic did not provide additional detail here on how much higher output token usage might be in typical workloads, and the company referenced an unreleased model, Claude Mythos, that outperformed Opus 4.7 in the cited benchmarks. As a result, users evaluating Opus 4.7 may weigh the reported capability gains against potential token consumption and the evolving competitive landscape reflected in the benchmark comparisons.