Grok 4.5 became selectable in GitHub Copilot on 28 July. On its own that is a routine changelog line.
Put it next to the rest of July and it is not routine at all: Copilot now carries models from OpenAI, Google, Anthropic and xAI inside one subscription.
The four-vendor position
July's model additions to Copilot ran as follows:
- 9 July — GPT-5.6 Sol, Terra and Luna (OpenAI)
- 21 July — Gemini 3.6 Flash (Google)
- 24 July — Claude Opus 5 (Anthropic)
- 28 July — Grok 4.5 (xAI)
Admins still hold the switch
As with Opus 5, Business and Enterprise administrators decide whether their developers ever see Grok 4.5. If it has not appeared in your model picker, check the Copilot model policy in your own admin console before assuming a rollout delay.
GitHub reinforced this the very next day with a separate change to default model enablement, which is covered in its own entry for 29 July.
Does more choice help?
Only if you use it. In practice most developers pick a model once and never touch the selector again. The value of four vendors is not daily switching — it is that a deprecation, an outage or a price rise at any one lab stops being your problem.
That is not hypothetical. Three days after this, GitHub announced August deprecations for two Gemini models.