Starting June 1st, GitHub Copilot will move to a usage-based billing model as GitHub Copilot supports more agentic and advanced workflows. In early May, you'll see a preview bill experience, giving visibility into projected costs before the transition. 👉 Read more about the upcoming change: https://lnkd.in/gA2vPT6i
Usage-based billing for AI coding assistants is a double-edged sword. On one hand, it aligns cost with value. On the other, it creates a psychological barrier — developers might start second-guessing every Copilot suggestion instead of using it freely as a thought partner. Will this actually reduce AI adoption among indie devs and small teams, or will it push them toward open-source alternatives like Codeium and Continue?
Santosh B.
1h
Is there any real value add with github copilot over direct claude subscriptions, if one was only using claude models on the copilot ? I was using copilot to avoid the heartburn I was getting with claude of hitting limits. On the flipside maybe it's better this way, I was getting addicted to building stuff, it felt like a video game ride you cleared round and onto next round
Saw this miles away, we all probably knew the tokenomics for GHCopilot didn't make sense at all (you could literally use up to 1-10 million+ of tokens in a single request, even considering subsidies, this is way too much compared to other coding agent plans), nor were they sustainable. Thanks for the generous limit while it lasted, GitHub :)
It’s becoming increasingly clear that Microsoft didn’t fully think through the pricing model for Copilot Pro when they first launched the product. The removal of Opus was the first red flag... The increasing infrastructure costs are understandable, but the recent changes point more to reactive adjustments than to a well-planned strategy. It’s not uncommon to see models take a position, then step it back - perhaps in an attempt to mimic human reasoning. But every one of those steps consumes tokens. If that’s the case, should users also be refunded when the model hallucinates or generates unnecessary back-and-forth?
Guess we're all being pivoted to a pricing model that is more aligned with the vendors need 💸 It's been a good ride though, so some more cost focus should be introduced in the market. Can't shake the feeling that we're experiencing the well-established "the first hit is free"-strategy 😅
That's it. We need a dislike button on Linkedin
Stefan Noll
2d
The removal of Opus from GitHub Pro was really a big disappointment. Will be Opus be back, and the key is how much the tokens will cost?
Victor Ike
2d
The enshitification of everything continues…
A usage based model where the vendor gets to control most of the usage rate is genuinely insane. If I ask a simple question & the model chooses to reply with War & Peace, that’s wholly to the benefit of the vendor and wholly at my expense. Beyond being compute starved, where is the incentive for concise or accurate output in this paradigm? To the accuracy point: if we’re being directly charged for every piece of output, when can we start talking about refunding credits for hallucinations?