Claude and Codex Now Available for Copilot Business and Pro Users

Multi-Agent Access Expands to Business and Pro Tiers

GitHub has expanded access to Anthropic's Claude and OpenAI's Codex as third-party coding agents within GitHub Copilot, now available to Copilot Business and Copilot Pro subscribers. This completes a phased rollout that began earlier in February 2026 with Copilot Enterprise and Pro+ users.

The expansion means that all paid Copilot tiers — Pro, Business, Enterprise, and Pro+ — now have access to the same multi-agent platform, removing the tier-based gating that previously restricted third-party agent access to higher-priced plans.

Unified Agent Platform

The three agents — GitHub's own Copilot agent, Anthropic's Claude, and OpenAI's Codex — run on a single shared platform with unified governance controls. This means that enterprise policies, content exclusion rules, and audit logging apply consistently across all three agents, regardless of which underlying model is powering the interaction.

Administrators do not need to configure separate policies for each agent. A single set of organization-level controls governs what all agents can access, what actions they can take, and how their activity is logged. This unified approach significantly reduces the administrative burden of managing a multi-agent development environment.

Shared Context and Copilot Memory

All three agents share the same context infrastructure, including repository context, conversation history within a session, and Copilot Memory. Copilot Memory is GitHub's persistent knowledge store that accumulates project-specific context over time — coding conventions, architectural decisions, team preferences, and other project knowledge that helps agents produce more relevant and consistent outputs.

Shared context means that switching between agents mid-task does not require re-explaining the project setup or re-establishing context. A developer can start a task with one agent, switch to another for a different phase of the work, and the second agent picks up with full awareness of what has already been done.

Agent Selection and Routing

Developers can explicitly choose which agent to use for a given task, or they can let the platform route their request to the most appropriate agent based on the nature of the task. The routing logic considers factors such as the type of work (code generation, code review, debugging, documentation), the programming language, and the complexity of the request.

Explicit agent selection is available through a model picker in all Copilot surfaces — VS Code, github.com, GitHub Mobile, and the CLI. The picker shows all available agents and their current status, making it straightforward to switch between them.

Availability

Claude and Codex are available immediately for all Copilot Pro and Business subscribers at no additional cost beyond the existing Copilot subscription. Usage of third-party agents counts against the same premium model usage limits that apply to other advanced Copilot features. Enterprise customers retain the ability to restrict which agents are available to their organization through the Enterprise AI Controls panel.