Which model does github copilot currently use?

[dubesar]

Select Topic Area

Question

Body

When i asked github copilot in chat, it says its using GPT4. Is it true? Has Github copilot started using GPT4? What is being used for code autocompletitions. I saw a similar question long time ago in this discussion forum but that stands irrelevant now as a lot of things might have changed.

[dubesar]

I am asking which model does it uses?

[spagu]

Uses GPT3 - with data back in 2021.

[grid-knight]

If you go to https://github.com/github-copilot/signup you will see it uses GPT-3.5 Turbo model to offer coding suggestions.

[spagu]

https://www.microsoft.com/en-us/store/b/copilotpro

was upgraded to gpt4

[grid-knight]

Yeah while not for Visual Studio Code, if I see that correctly.

[nicocossiom]

https://www.microsoft.com/en-us/store/b/copilotpro

was upgraded to gpt4

That's not the same copilot as GitHub's

[github-actions[bot]]

🕒 Discussion Activity Reminder 🕒

This Discussion has been labeled as dormant by an automated system for having no activity in the last 60 days. Please consider one the following actions:

1️⃣ Close as Out of Date: If the topic is no longer relevant, close the Discussion as out of date at the bottom of the page.

2️⃣ Provide More Information: Share additional details or context — or let the community know if you've found a solution on your own.

3️⃣ Mark a Reply as Answer: If your question has been answered by a reply, mark the most helpful reply as the solution.

Note: This dormant notification will only apply to Discussions with the Question label. To learn more, see our recent announcement.

Thank you for helping bring this Discussion to a resolution! 💬

[Kevin-Pro-Codes]

is it actually using gpt 4?

[edsonmedina]

It should be gpt-4o since 1.92 (July 2024):

https://code.visualstudio.com/updates/v1_92#_github-copilot

[ayim]

At time of writing (Sept 2024), It's using 4o.
VSCode: https://code.visualstudio.com/updates/v1_92#_github-copilot
VS: C:\Users<username>\AppData\Local\Temp\VSGitHubCopilotLogs - you'll see the response is generated using 4o.

The GitHub team are in the middle of testing and prototyping improvements using o1, blog post on that here:
https://github.blog/news-insights/product-news/openai-o1-in-github-copilot/

[g0t4]

35-turbo for completions (stumbled on an error in the copilot logs with vscode insiders + pre-release of copilot w/ o1 support), note this is not the chat panel and I was able to reproduce the error after clearing the logs so definitely is for completions.

[g0t4]

confirmed with mitmproxy ... the response specifies the model (on Ctrl+Enter) and they're all gpt-35-turbo, the request does not ask for a model so IIAC they could be changing it based on the request... I'll monitor traffic for a while and see if I notice any other models

[ayim]

;)

We're always working on improvements, the most recent one that shipped for VS was using related files (even if they're closed) and awareness of valid types/methods to reduce hallucination: https://devblogs.microsoft.com/dotnet/improving-github-copilot-completions-in-visual-studio-for-csharp-developers/

[piskov]

I’m more of a Jetbrains Rider guy myself: hope this will extend to Copilot extension in Rider as well :-)

[nvmnghia]

so it's still 3.5 or 4o now? thx