Anthropic released Claude Opus 5 on 24 July. GitHub made it selectable inside Copilot the same day. For anyone who has been paying for a second coding tool purely to reach Anthropic’s strongest model, that is the more consequential of the two announcements.
The reason is simple: until now, model choice was the clearest argument for leaving Copilot. That argument just got weaker.
What GitHub actually shipped
The changelog entry is short and specific. Claude Opus 5 is available to Copilot Pro+, Max, Business and Enterprise — note that plain Pro is not on the list. It can be selected from an unusually long list of places:
- Visual Studio Code, Visual Studio, JetBrains, Xcode and Eclipse
- The Copilot CLI and the Copilot cloud agent
- github.com, the Copilot app, and GitHub Mobile on iOS and Android
Two things that will trip teams up
First, the rollout is gradual. GitHub’s own wording is "check back soon if you don’t see it yet", so an absent model picker is not a bug to raise with support on day one.
Second — and this is the one that generates the support ticket — Business and Enterprise administrators have to enable the Claude Opus 5 policy in Copilot settings before anybody on the plan can choose it. If your team is on Enterprise and the model has not appeared, the blocker is almost certainly an unticked box in your own admin console rather than anything on GitHub’s side.
One thing the announcement does not say: whether Opus 5 carries a premium request multiplier. GitHub’s post is silent on it, so treat per-request cost as unconfirmed until it appears in the billing documentation rather than assuming parity with cheaper models.
Copilot is now a three-vendor assistant
Read this alongside the rest of July and a pattern emerges. GPT-5.6 Sol, Terra and Luna landed in Copilot on 9 July. Gemini 3.6 Flash followed on 21 July. Opus 5 arrived on 24 July. Inside a single subscription, a developer can now route work to OpenAI, Google or Anthropic models.
That is the same bet Cursor made three days earlier with Cursor Router, which picks a model per request across Intelligence, Balance and Cost modes. The difference is who does the choosing: Cursor automates it, Copilot still asks you to pick. Neither approach is obviously better, but both vendors have concluded that being tied to one model house is a liability.
Does this change which tool you should buy?
For teams already standardised on GitHub, it removes a real reason to add a second subscription. Copilot scores 8.6 on our scale against Cursor’s 9.0, and much of that gap has been about multi-file agent work rather than raw model access — so closing the model gap narrows the case for switching without eliminating it.
For anyone starting a project from nothing, neither is the right first stop; that is what a build agent like Emergent is for. And if the work is understanding an unfamiliar codebase before writing anything, Claude on its own — at 9.1, the highest score in our directory for this kind of work — is still the better instrument than either editor.
The honest summary: this is a procurement story more than a capability story. The model was already available to anyone willing to pay for a second tool. What changed is that fewer teams now need to.