Claude Opus 5 enters long-running coding work in GitHub Copilot

What was announced

Anthropic introduced Claude Opus 5 through its official newsroom on July 24. According to Anthropic, Opus 5 is the new default model for Claude Max and is offered as the strongest model in Claude Pro.

On the same day, GitHub also announced in its official Changelog that Claude Opus 5 has arrived in GitHub Copilot. That means this news combines Anthropic’s standalone model release with GitHub Copilot availability on the same date.

Where it can be used in Copilot

GitHub said Opus 5 is available to Copilot Pro+, Max, Business, and Enterprise users. The model picker is being rolled out across Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot cloud agent, Copilot app, github.com, GitHub Mobile, JetBrains, Xcode, and Eclipse.

However, this does not mean it will appear immediately for every account. GitHub said the rollout is gradual, and Business and Enterprise administrators must enable the Claude Opus 5 policy in Copilot settings.

Why it is viewed as a model for long-running coding work

The use cases GitHub emphasized are not short Q&A sessions, but complex, long-running coding tasks. For example, GitHub said the model showed strengths in autonomous code changes, regression verification, and tasks that require coordinating several tools together.

That points to a broader flow than “suggest one line of code.” In agentic development flows where a model reads an issue, finds the change points, runs tests or verification, and leaves the result as a draft pull request, model selection becomes a more direct operational variable.

Pricing and usage also matter

Anthropic lists Opus 5 API pricing at $5 per million input tokens and $25 per million output tokens. It described this as the same base pricing as Opus 4.8.

On GitHub’s side, Opus 5 is billed through usage-based pricing based on the provider API list price. So enabling a high-performance model in Copilot is both a performance choice and a cost-management choice.

Why security limits were mentioned

GitHub said Opus 5 includes enhanced safeguards for high-risk cyber content, and that some security-adjacent requests may be blocked. If a request is blocked, GitHub advises adding clearer benign context or choosing another model.

Anthropic’s official post also separately explains Opus 5 safeguards related to cyber and biological domains. For readers, the signal is that the stronger the coding model, the more important it is to check permissions, logs, review practices, and blocking policies together.

How this differs from the earlier Opus 4.8 news

Game Sunakku previously covered Claude Opus 4.8’s Copilot integration and fast mode flow. This Opus 5 update is an extension of that story, but it differs in that a new model generation, multi-client Copilot rollout, and usage-based cost guidance were all announced together.

In short, this change is less about “a new model is out” and more about AI coding tools entering a more detailed stage where teams choose which model to use for which task and how much to spend.