Questions about the model of GitHub Copilot Pro

[ytooyama]

🏷️ Discussion Type

Question

💬 Feature/Topic Area

Copilot in GitHub

Body

I have a contract with GitHub Copilot Pro.
The biggest advantage of GitHub Copilot was that it could use various models, but the models are disappearing more and more.

Why can I use GPT-4o in GitHub Copilot CLI, but I can't use GPT-4o in VSCode?
On the other hand, Gemini cannot be used in GitHub Copilot CLI.
Why is there a difference between the model provided by CLI and Code?
There will be many changes on June 1st, but is there a possibility that the number of models will increase?

I'm not good at English, so I'm sorry for my awkward English.

[Arqam0919]

Here is the short answer to your questions:Why GPT-4o works in CLI but not VS Code: This is a bug. GPT-4o is available in VS Code. To fix it, go to your GitHub account settings online, make sure all models are enabled, and update your VS Code Copilot extension to the latest version.Why Gemini is missing from CLI: The CLI tool is built on older, separate code. GitHub has not updated the CLI to support Google Gemini yet.Will models increase on June 1st? No, the total number of models will likely decrease at first. On June 1st, GitHub shifts to an "AI Credits" system, and some expensive models (like Claude 3 Opus) are actually being removed from standard plans. However, this new credit system makes it easier for GitHub to add new models in the future.

[ytooyama]

Thank you for your reply.
I think the recent Visual Studio Code was a bundle of Copilot extensions.
Therefore, extensions cannot be updated.

It may be a temporary bug.
I'll try to restart it or try it as a login & logout to see if it can be solved.

[FahimAhamed0602]

GPT-4o is currently supported in VS Code. If it is not appearing, your extension is likely out of sync with your account permissions. To fix this, verify your model access permissions in your GitHub web settings, then completely sign out and sign back into your GitHub account inside VS Code to force a refresh.

The reason you see different models in the CLI versus VS Code comes down to how they are built. The Copilot CLI and the VS Code extension run on completely different underlying architectures. GitHub prioritizes the code editor extensions, so they receive the newest AI model integrations first. The CLI tool operates on older infrastructure that has simply not been updated to support newer models like Gemini yet.

The changes coming on June 1st involve GitHub transitioning Copilot to a usage-based "AI Credits" billing model. Instead of a flat allowance of premium requests, your usage will be calculated based on token consumption. Because running massive language models is expensive, the standard Copilot Pro plan is removing access to high-tier models like Claude 3 Opus, restricting them to the more expensive Pro+ tier. While you may lose access to the heaviest models on the standard plan in the short term, transitioning to a token-based system gives GitHub the financial sustainability to plug in a wider variety of new models in the future.

[ytooyama]

I've already tried to sign out of my GitHub account completely and then sign in again, but that didn't solve the problem.
There is no GPT-4o in my VSCode.

By the way, will GPT-5.5 not be published in the Pro subscription for the same reason as Opus,? I don't understand it well because it's not written honestly in the document.

[FahimAhamed0602]

1. Troubleshooting GPT-4o in VS Code

Update Extensions: Ensure that both the GitHub Copilot AND GitHub Copilot Chat extensions are updated to their absolute latest versions. The base Copilot extension and the Chat extension update separately.

Switch to Pre-Release: Sometimes new models are gated behind the pre-release version of the extension. Go to your VS Code Extensions tab, find GitHub Copilot Chat, and click "Switch to Pre-Release Version" to see if that forces GPT-4o to appear.

Update VS Code: Ensure your actual VS Code editor is on the latest version, as older builds can sometimes block extension updates.

Manual Selection: In the Copilot Chat panel, check the model dropdown (usually located near the chat input box or at the top of the panel) to see if you can manually toggle it to GPT-4o.

2. Regarding GPT-5.5 and Future Models
   You are completely right that the official documentation is vague right now! GitHub has not made any official announcements regarding GPT-5.5 (as that model hasn't been officially released yet).

However, looking at the precedent they are setting with Claude 3 Opus, it is highly likely that future "ultra-premium" or highly complex models will follow the same pattern. As AI models get larger and more expensive to run, standard Pro subscriptions will likely stick to highly capable but more cost-effective models, while the heaviest computational models will be restricted to higher tiers or consume more of the upcoming "AI Credits." The documentation is likely unclear because GitHub is still actively figuring out the pricing structure for these unreleased models.