OnMicro.AI

Blog

Custom GPT Alternatives for Educators: What to Use After December 11

Plugins are built for your workspace. Your students need something built for the classroom.

John Swope

Custom GPTs were never OpenAI’s biggest hit. But for educators, they quietly filled a niche by being easy to build and easy to share. A faculty member could build a guided tutor on a Sunday night and send students a link on Monday morning, with no developer, no IT ticket, and no procurement.

That ends on December 11, 2026, when OpenAI retires Custom GPTs and moves creators to plugins. I wrote earlier about what educators will lose when Custom GPTs go away. This post is the practical follow-up: what to use instead.

Here is the TL;DR:

What happens to Custom GPTs on December 11?

Custom GPTs stop running on December 11, 2026, and OpenAI’s replacement is plugins. According to OpenAI’s retirement and migration FAQ, your instructions and reference files can be migrated into a plugin, but custom actions do not transfer automatically. Enterprise workspaces with an approved deferral get until February 11, 2027.

The part that matters most for educators is sharing. A migrated plugin starts private, and the original GPT’s sharing settings do not carry over. Personal ChatGPT accounts can no longer create or publish new GPTs at all. When I migrated one of my own tutoring GPTs on a personal account, there was no option to share it with students. The only official way to hand it to someone else who isn’t in your workspace is to share a zip file that another user can upload into their plugins directory.

To many users, plugins are a seamless replacement. But plugins are built for people working inside the same workspace, and most students are not in your workspace.

What should educators look for in a Custom GPT alternative?

The right alternative depends on who will use it. For anything student-facing, I’d judge every option on six things:

  1. Student access without a paid account — Can a student use it without buying their own AI subscription? In a 2025 survey of 1,000 UK students run by the AI vendor ICS.AI, 77% used free AI tools only, and 45% felt at least slightly disadvantaged compared with peers who pay.
  2. Shareable by link — Can you hand it out the way you handed out a Custom GPT?
  3. LMS integration — Can it live inside Canvas, Moodle, or Google Classroom, with students signed in and completion visible in the gradebook?
  4. Conversation visibility — Can you see where students struggled, not just whether they showed up?
  5. Data protection — Is it covered by your institution’s agreements, with student data kept out of model training?
  6. Cost — Is it included in something your campus already pays for?

The 6 Custom GPT alternatives, compared

Tool Best for Student access Share by link LMS integration See student conversations Cost
ChatGPT plugins Faculty using AI for their own work Students must be in your ChatGPT workspace Workspace only None documented Not documented Your ChatGPT plan
Gemini Gems Google Workspace campuses School Google account Yes, inside Google Workspace Google Classroom Usage analytics for notebooks (AI Pro for Education) Included in Workspace for Education; Pro is an add-on
Claude Projects Campuses with Claude for Education Institution’s Claude plan Shared with project members None documented No — chats are private unless the student shares them Institutional plan
Microsoft Copilot agents Microsoft 365 campuses School Microsoft 365 account Inside your Microsoft 365 tenant None documented IT-level usage reporting Copilot Chat included with A1/A3/A5; agents on institutional data are pay-as-you-go
MagicSchool Student Rooms K-12 classrooms Teacher-created Rooms Room join Canvas, Schoology, Google Classroom (Enterprise) Yes, Room monitoring and insights LMS integrations require Enterprise
OnMicro Student-facing exercises in higher ed No AI account needed Yes, direct link with site restrictions Any LMS that supports LTI 1.3 (That’s most of them) Yes, conversation analytics $0-$79/m per creator. LTI Connection requires Enterprise.

Comparison based on each vendor’s published documentation as of November 2026. Features change quickly; check the linked docs before you decide.

ChatGPT plugins: best for you, not your students

Plugins are simultaneously a better tool for the future of OpenAI and a worse tool for the classroom. You can stack them, connect them to more apps, and call one in any chat. For your own grading prep or course design, that’s an upgrade. But OpenAI’s FAQ describes sharing in workspace terms, so students outside your workspace are left out.

Gemini Gems: the strongest pick if your campus runs Google

If your institution is on Google Workspace for Education, Gems are the closest thing to a Custom GPT that students can actually use. Teachers can create Gems and assign them in Google Classroom, and Gemini for Education is included in every Workspace for Education edition at no added cost. The catch: your admin must turn on Gem sharing, and Google’s own help page says teachers can’t view the notebooks students create.

Claude Projects: strong for guided learning, limited visibility

On campuses with a Claude for Education agreement, faculty can build a Project with instructions and course files and share it with students. Learning mode, which guides students’ reasoning instead of handing them answers, runs inside Projects. The limit is visibility: LSE’s faculty guide notes that chats in a Project stay private unless the student shares them.

Microsoft Copilot agents: the default for Microsoft 365 campuses

If your school runs on Microsoft 365, Copilot Chat is available at no added cost with A1, A3, and A5 licenses, including for higher-ed students. Agents in Copilot Chat are free when grounded in web data and pay-as-you-go when they use institutional data. IT can monitor usage centrally, but this is an IT-managed tool, not something a faculty member spins up and links in a syllabus.

MagicSchool Student Rooms: built for K-12

MagicSchool’s Student Rooms give teachers a moderated space where students use AI tools the teacher chooses, with monitoring built in. Its integrations cover Canvas, Schoology, and Google Classroom for Enterprise customers. It’s an excellent fit for K-12. For a college course with a specific article and a multi-phase tutoring flow, it’s less of a match.

OnMicro: built for student-facing exercises

Disclosure: I built OnMicro, so weigh this section accordingly.

I built OnMicro to fill this exact gap. You paste in your Custom GPT’s instructions and files, and it turns them into a structured, multi-step app. Students open it with a link, no AI account needed, or launch it from Canvas through LTI 1.3 so completion shows up in the gradebook. You can then read student conversations to see where the class is struggling. It is not a general-purpose chatbot, and it is not trying to be.

Which Custom GPT alternative should you choose?

It depends on who the GPT is for. Here’s how I’d sort it:

  1. Only you use it — Migrate to a ChatGPT plugin and move on.
  2. Students use it, and your campus licenses a platform — Start with Gemini Gems, Copilot agents, or Claude Projects, whichever your institution already pays for.
  3. Students use it, and you need link sharing, gradebook integration, or conversation review — Use a purpose-built tool like OnMicro, or ask your teaching and learning center what your campus supports.

The Result:

The Custom GPT retirement is a move toward the workspace and away from the classroom. That is a reasonable product decision for OpenAI. It is also clearly a gap for educators, because educators are not developers and students are not workspace members.

Before December 11:

  1. Inventory your GPTs — List each one and who uses it: you, colleagues, or students.
  2. Migrate everything to a plugin anyway — It preserves your instructions and files, even if students won’t use the plugin.
  3. Give student-facing GPTs a new home — Pick from the table above, and test it with a few students before next term.

If you want the step-by-step, I walk through moving a tutoring GPT into a student-facing app here: How to Migrate a Custom GPT for Students Before December 11.

FAQ

What replaces Custom GPTs?

OpenAI is replacing Custom GPTs with plugins, which bundle instructions, reference files, and connected apps. For student-facing use, educators are turning to Gemini Gems, Microsoft Copilot agents, Claude Projects, MagicSchool, or purpose-built tools like OnMicro.

Can students use a Custom GPT alternative without a ChatGPT account?

Yes. Gems, Copilot agents, and Claude Projects work through your institution’s Google, Microsoft, or Claude accounts. Tools like OnMicro work through a direct link or an LMS launch, with no AI account at all.

Will my Custom GPT still work after December 11, 2026?

No. Custom GPTs stop running on December 11, 2026, unless you’re in an Enterprise workspace with an approved deferral, which runs to February 11, 2027. Past conversations remain viewable.

Can I see what students ask an AI tutor?

It depends on the tool. Claude Projects keeps student chats private unless shared, Copilot gives IT usage reporting, and MagicSchool and OnMicro let instructors review student activity.


AI Disclaimer: AI was used in the writing of this article from my outline and goals. I made final edits.

John Swope is the founder of Curricu.me for AI enablment for Education, and OnMicro.AI