GitHub widens Copilot model support

GitHub Copilot is no longer documented as a one-model assistant. Its docs now present Copilot as a layer that can route across a growing list of models from OpenAI, Anthropic, Google and xAI. (docs.github.com) The current supported-models page lists GPT-4.1, GPT-5 mini, GPT-5.2, GPT-5.2-Codex, GPT-5.3-Codex, GPT-5.4, GPT-5.4 mini, GPT-5.4 nano, Claude Haiku 4.5, Claude Sonnet 4, Claude Sonnet 4.5, Claude Sonnet 4.6, Claude Opus 4.5, Claude Opus 4.6, Claude Opus 4.7, Gemini 2.5 Pro, Gemini 3 Flash, Gemini 3.1 Pro and Grok Code Fast 1. (docs.github.com) That same page also lists fine-tuned models such as Raptor mini and Goldeneye, and it breaks availability out by Agent mode, Ask mode and Edit mode instead of treating Copilot as one fixed backend. (docs.github.com) GitHub’s own guidance says model availability can change over time. In supported IDEs, “Auto” can pick the best model based on availability, and users can still override that choice manually. (docs.github.com 1) (docs.github.com 2) The practical change is that Copilot behavior now depends on plan, product surface and routing policy. GitHub says Auto model selection is available in Copilot Chat for all plans, and in Copilot cloud agent for Copilot Pro and Pro+ plans. (docs.github.com) GitHub also ties model choice directly to cost. Its billing docs say Copilot Chat, Copilot CLI, code review, cloud agent and Spaces can consume premium requests, with usage multiplied by the model’s rate. (docs.github.com) Auto routing is meant to smooth that complexity rather than remove it. GitHub says Auto chooses models based on real-time system health and model performance, excludes models above a 1x multiplier, and gives paid plans a 10% multiplier discount in Copilot Chat. (docs.github.com) (github.blog) The docs also show how quickly the lineup is changing. GitHub’s retirement table says GPT-5.1 was retired on April 15, 2026, while GPT-5.1-Codex, GPT-5.1-Codex-Max and GPT-5.1-Codex-Mini were retired on April 1, 2026. (docs.github.com) GitHub is still adding models at the same time. Its April 2026 changelog says GPT-5.5 began rolling out to GitHub Copilot on April 24, 2026, even though the comparison docs already mention GPT-5.5 as a deep-reasoning option. (github.blog) (docs.github.com) The company has widened the architecture beyond GitHub-hosted routing too. On April 7, 2026, GitHub said Copilot CLI could use Azure OpenAI, Anthropic, OpenAI-compatible endpoints, or local runtimes such as Ollama, vLLM and Foundry Local. (github.blog) GitHub’s Copilot product now looks less like a single assistant and more like a traffic controller: one interface, many models, shifting defaults, and a billing meter attached to the route it takes. (docs.github.com 1) (docs.github.com 2)