Github Copilot
Your AI pair programmer
Start new thread

GitHub Copilot is now usage-based. What's your team doing about it?

by•

Starting June 1st, moved to a usage-based billing model.

Question for engineering leaders after this change:

What's your team actually doing about it?

591 views

Add a comment

Replies

Best

The shift to usage based pricing makes AI costs much more visible. A hybrid approach feels practical , teams get flexibility while keeping spend predictable as usage grows.

Cancelled plan this month. Gonna try Cursor next month.

no longer using ? have you considered AI-enabled extensions like and ?

VS Code is good for sure. I mean cursor is also popular lately. I wanna give the composer a try see if it works along with me😂

curious to have your thoughts about it! see this thread:

 Sure thing, gotta try it next month to have some thoughts. coz we already paid for github copilot this month😂

Canceled plan, using china models for a while.

curious what are your preferred models? ? ? else?

  Glm and qwen but models are less relevant read my article here

Microsoft has been cutting and cutting what Copilot can do on each of their oh-so-confusing myriad copilot plans. Things I could do last week, I can no longer do. I am so tired of Microsoft's bait-n-switch. I was a beta tester long before they started grouping beta testers and long before the "Insiders" programs began. I'm tired of it. I started slowly switchiing my preferred products a few years ago. It was time for a upgrade to my Surface Pro. I bought a MacBook Air. It is getting close to time to renew my Microsoft 365 subscription. Protonm is an excellent, much more secure end-to-end encryption office suite. Microsoft has gotten too big for its britches!

 what alternative to would you recommend?

cancelled plan next month. already try Claude Code, feel great

feels like still vendor locked-in?

 Yeah, the vendor lock-in feeling is pretty real in this space. Even when switching models or tools, the GitHub ecosystem (repos, PRs, workflows) still pulls you back in. Curious what specific parts feel most locked-in for you?

As a solo founder, I’m becoming much more conscious of AI ROI. I don’t mind paying for usage if it saves meaningful engineering time, but it definitely changes how casually people will use agents.

I'm not using Copilot, but invoices are coming to Claude starting next week as well. I built a tool (I launched here a week or so ago) to support cross-session memory querying. Now it's even more of an imperative to use tools like this. The past month, I've been on a mission to fine-tune my agents to keep them under my monthly budget. It's challenging for sure. I've moved some agents to cheaper or local models, moved as many deterministic things as possible to scripts, etc.

We saw this coming but it still stings. The predictability of a flat monthly fee was the whole reason we picked Copilot for the team in the first place. Now we're back to having conversations about "who's using too many completions" which is exactly the kind of overhead we were trying to avoid by paying for a tool.

For now we're staying on Copilot but watching usage closely for the first month to see where the bill actually lands. If it spikes past what we were paying before, we'll probably move half the team to Claude Code or Cursor and keep Copilot only for the heaviest users. The irony is that a tool meant to save developer time is now creating admin work to manage its own costs.

usage-based sounds fair until nobody on the team is actually tracking their usage. the bill just shows up and then it's a fire drill. hybrid makes the most sense to me... cap the heavy stuff, route simpler tasks to cheaper alternatives. single vendor lock-in was already risky before they changed the pricing

hybrid is leading the way

Usage-based is the part that bites: your cap is now premium requests (300/mo on Pro, 1,500 on Pro+) with overage billing, so "unlimited" quietly became a meter. I got tired of every provider changing these terms without telling anyone, so I built LimitWatch to track current limits and how they drift over time, every number linked to its source. Copilot's on there:

123
Next
Last