What AI model does it use?

What AI model does GitHub Copilot use?

Ask which model Copilot uses and the true answer is a table. GitHub's supported models page, which I read on September 25, 2026, lists 32 models, all generally available, from six companies. Which of them you actually get depends on your plan and on whether you've left Copilot on Auto.

Who makes the 32 models

OpenAI has the most, twelve, running from GPT-5 mini up to GPT-6 Sol and GPT-6 Astra. Anthropic has ten, from Claude Haiku 4.5 up to Claude Opus 5.5 and Claude Fable 5.1. Google has four Gemini Flash models, xAI three Grok models, and Moonshot AI two Kimi models.

The sixth is Microsoft itself, with MAI-Code-1.1-Flash, which replaced MAI-Code-1-Flash on September 10. GitHub belongs to Microsoft, so this is the one model in the picker that comes from inside the family, and it's available on every plan.

Free and Student: Auto only

If you're on Copilot Free, you don't choose. The docs are direct about it: "Copilot Student and Copilot Free users have access to models through auto model selection only."

Auto routes each request based on how hard the task looks and on live system health. Its pool on September 25 had 17 models, including MAI-Code-1.1-Flash, GPT-5.4 through GPT-6 Sol, and Claude Haiku 4.5, Sonnet 5 and Opus 5.5. Paid users who leave Auto on get a 10% discount on model costs, which is GitHub nudging people toward it.

Auto can also serve unnamed test models to individual-plan users (it's a line that's easy to scroll past): "Evaluation models may appear in the product with codenames rather than official model or provider names." You can opt out in your settings.

What Pro can and can't pick

Copilot Pro, the $10 plan, gets a long menu: Claude Haiku 4.5, Sonnet 5 and Sonnet 4.6, all the Gemini Flash models, several GPT-5 models, GPT-6 Luna, Grok, Kimi and MAI-Code-1.1-Flash. It doesn't get the biggest ones. The Opus models, Fable, GPT-5.5, GPT-5.6 Sol, GPT-6 Sol and GPT-6 Astra are marked "Not included" for Pro and need Pro+, Max, Business or Enterprise.

The fallback: GPT-5.3-Codex

On May 17, 2026, GitHub made GPT-5.3-Codex "the base model for all Copilot Business and Copilot Enterprise organizations, replacing GPT-4.1." It's also GitHub's first long-term support model, promised through February 4, 2027, and it's the last resort for Auto: "GPT-5.3-Codex will be used in the event no other models are available."

Multipliers are gone for most people

A lot of what ranks for Copilot model questions explains premium request multipliers. Those stopped applying to most plans on June 1, 2026, when every plan moved to GitHub AI Credits billed by tokens used (one credit is a cent). Only annual Pro and Pro+ subscribers who stayed on the old billing still count premium requests.

What GitHub doesn't say

GitHub names no default chat model for individual plans, since Auto is the default path. It doesn't say which model does code review; the pricing page says that model "is selected automatically and is not disclosed". And the list moves fast: Claude Opus 4.7, Gemini 3.5 Flash, Gemini 3.6 Flash and Kimi K2.7 Code are all due to retire on October 2, 2026.