GitHub Copilot Drops Claude Opus 4.1 and GPT-5 in Latest AI Model Purge

GitHub has pulled the plug on three AI models powering its Copilot coding assistant, forcing developers to migrate to newer alternatives as the platform continues its aggressive model refresh cycle.

The deprecations, effective February 17, 2026, affect Claude Opus 4.1, GPT-5, and GPT-5-Codex across all Copilot experiences including Chat, inline edits, agent modes, and code completions. GitHub is pointing users toward Claude Opus 4.6, GPT-5.2, and GPT-5.2-Codex as replacements.

Part of a Broader AI Housecleaning

This move follows OpenAI’s own retirement of several ChatGPT models just four days earlier on February 13. The company pulled GPT-5, GPT-4o, GPT-4.1, and OpenAI o4-mini from the ChatGPT interface, citing remarkably low adoption—only about 0.1% of daily users were still touching GPT-4o.

GPT-4o’s journey has been particularly rocky. OpenAI briefly withdrew it during the GPT-5 rollout in August 2025, then brought it back after user complaints, only to kill it again this month. Criticism over “sycophantic” behavior—the model’s tendency to tell users what they wanted to hear—apparently didn’t help its case.

Anthropic has been running a parallel cleanup. Claude 3.5 Sonnet and its v2 variant hit their shutdown date on February 19, while Claude 3.5 Haiku faces a July 5 cutoff after being deprecated in early January.

What Admins Need to Do

Enterprise customers can’t just sit this one out. Copilot Enterprise administrators need to enable access to the replacement models through their model policies in Copilot settings. Once configured, the new models will appear in the Chat model selector in VS Code and on github.com.

The deprecated models will disappear automatically—no manual removal required. But if your team’s workflows or integrations reference specific model versions, those will break without updates.

GitHub Enterprise customers with concerns should contact their account managers. For everyone else, the documentation on supported models has been updated to reflect the changes.

The rapid deprecation cycle highlights an uncomfortable reality of building on top of AI infrastructure: the foundation keeps shifting. Developers who hardcoded model preferences into their workflows are learning that lesson again.

Image source: Shutterstock