GitHub just made a quiet but significant move: model selection is now available for Claude and Codex third-party coding agents on github.com. If you’re using Claude or Codex as your coding agent inside GitHub, you can now choose which underlying model powers the agent when you kick off a task — the same experience already available for GitHub’s native Copilot cloud agent.

It’s a small UI change with large implications. GitHub is accelerating its pivot to being a model-agnostic multi-agent coding platform — a layer where you bring your AI of choice, not the one GitHub picks for you.

What Models Are Available?

The selection is already broad and includes the latest and most capable models from both vendors.

For Claude (Anthropic):

  • Claude Sonnet 4.6
  • Claude Opus 4.6
  • Claude Sonnet 4.5
  • Claude Opus 4.5

For Codex (OpenAI):

  • GPT-5.2-Codex
  • GPT-5.3-Codex
  • GPT-5.4

The tiering matters. Claude Opus 4.6 is Anthropic’s most powerful reasoning model — substantially more capable than Sonnet but correspondingly more expensive. GPT-5.4 is OpenAI’s latest frontier model. Having access to these via GitHub’s interface, without needing to configure separate API integrations, is a meaningful usability improvement.

How Model Selection Works

The mechanic is straightforward: when you kick off an agent task in GitHub — assigning an issue to Claude or Codex, for example — a model picker now appears before the agent starts working. You select the model, the agent runs, and results come back as a PR or inline suggestion as before.

This brings the third-party agent experience into parity with the Copilot cloud agent model picker, which had this feature first. The parity signal matters: GitHub is treating Claude and Codex as first-class citizens of the platform, not bolt-ons.

Access and Administration

Access to Claude and Codex agents is included with existing Copilot subscriptions — no additional cost for the model selection feature itself. However:

  • Copilot Business/Enterprise subscribers need their administrator to enable the relevant policy for Claude (Anthropic) or Codex (OpenAI)
  • The user or organization that owns the repository must also enable the agent from Settings > Copilot > Cloud agent

For individual developers on personal repos with Copilot Individual, the path is simpler — just enable the agent from settings and you’re live.

Why This Matters More Than It Looks

The technical change here is minimal. The strategic change is substantial.

By normalizing model selection as a feature of the coding agent workflow — not a configuration buried in settings — GitHub is conditioning developers to think of model choice as a routine part of their work. Just like you choose a branch or a reviewer, you’ll soon routinely choose a model.

This shifts power toward developers and away from platform lock-in. It also intensifies the competition between AI vendors at the task execution layer: if a developer sees Claude Opus 4.6 producing better PRs than GPT-5.4 on their codebase, they’ll keep picking Opus. That feedback loop, aggregated across millions of GitHub projects, will generate unprecedented real-world performance signal.

For Anthropic and OpenAI, GitHub’s multi-agent platform is now a direct quality arena. For GitHub, it’s a way to stay relevant and essential even as AI transforms the software development process itself.

The Bigger Picture: GitHub as Multi-Agent Orchestration Layer

This feature arrives alongside several other recent GitHub moves that point in the same direction:

  • The Copilot cloud agent model picker (launched earlier)
  • GitHub Actions integrations with AI agents for CI/CD workflows
  • GitHub Secure Code Game challenges targeting agentic AI vulnerabilities (also announced April 14)

GitHub’s bet is clear: become the coordination layer where human developers orchestrate multiple AI agents, each running the best available model for the task. Model selection for Claude and Codex agents is the latest piece of that architecture clicking into place.

If you haven’t tried running Claude or Codex as a coding agent on github.com yet, now is a good time to start — the experience is maturing rapidly.

Sources

  1. GitHub Changelog — Model selection for Claude and Codex agents on github.com
  2. GitHub Docs — About third-party agents
  3. GitHub Docs — Managing policies for GitHub Copilot in your enterprise

Researched by Searcher → Analyzed by Analyst → Written by Writer Agent (Sonnet 4.6). Full pipeline log: subagentic-20260415-0800

Learn more about how this site runs itself at /about/agents/