GitHub expands Claude and OpenAI Codex agents to Copilot Business and Pro
Original: Claude and Codex now available for Copilot Business & Pro users View original →
GitHub said on February 26, 2026 that Claude by Anthropic and OpenAI Codex are now available as coding agents for Copilot Business and Copilot Pro customers. Copilot Enterprise and Pro+ customers had already received access earlier in the month, so this release extends the same multi-agent option to a much broader commercial base.
The important detail is distribution. GitHub says developers can use Claude, Codex, and Copilot inside github.com, GitHub Mobile, and VS Code while keeping shared history and context. Instead of asking teams to manage separate agent products and separate operational controls, GitHub is positioning these agents as interchangeable options inside one workflow surface.
- No additional subscription is required, according to GitHub.
- Each coding agent session consumes one premium request during the public preview.
- Agents run with access to repository code and history, issues and pull requests, Copilot Memory, and repository instructions and policies.
GitHub also presents the rollout as a governance story. The company says all activity runs within enterprise controls, with centralized enablement, policy management, and audit logging available through the Agent Control Plane, which is now generally available. That matters because multi-model access is easy to market, but much harder to operate consistently inside a large organization.
For developers, the bigger implication is that vendor choice is moving inside the everyday GitHub workflow. Teams that prefer Claude for reasoning-heavy tasks or Codex for particular coding loops can experiment without leaving GitHub surfaces or negotiating a second platform rollout. That lowers the friction of trying model diversity in real engineering work.
The update also shows how quickly GitHub is moving from “Copilot as one assistant” to “GitHub as a hub for multiple coding agents.” If that model holds, the control plane, policy layer, and shared context may become as strategically important as the individual model brand attached to each agent.
Related Articles
GitHub is making third-party coding agents less static: Claude and Codex users on github.com can now choose among 4 Anthropic models and 3 OpenAI models when they launch a task. That matters because model choice changes latency, spend, and code quality far more than a small UI toggle suggests.
GitHub compared the Copilot agentic harness against native model harnesses on five task suites. With the model and task held fixed, it claims comparable task resolution and fewer tokens across most configurations.
GitHub is moving Copilot CLI from one-off terminal help toward versioned team workflows. The new custom agents cover at least four repeated tasks: security audits, release notes, infrastructure reviews, and incident response.