I only use Copilot for code completion, the other features are insufferably bad, at least in my IDE. I’m beginning to feel like I’m seriously overpaying. Does MS not update their completion model, or is it just my impression that suggestion quality has been stagnant for literally years? Do you guys know any cloud based code completion services with performance superior to Copilot?
Does Cody have an agent mode? I have the free vesion and it doesn't seem to. IMO coding assistant without agent mode is useless, it's useful just only for asking questions.
Copilot used to be much worse than Cursor but they have agent mode and MCP now.
Also I use Cody as a backup because I thought their ollama support means I can code on the plane, but when I actually tried, it stills requiring SourceGraph access and failed.
I also only use the code completion. I recently started using the Continue extension in VSCode running local models. They get the job done and it's free.
It is unfortunately also abandoned. The creator got an offer from Cursor to integrate it last Sepmtember and abandoned it, seems likely it was part of the contract. By the looks of it he also didn't refund subscribers, not a good look.
I've had problems with pylance and auto complete on Cursor for months for (what I think is) the same reason - so I'm not sure the crackdown has only started in the last couple days
i was in previous iterations of products ms doing eee/enshittification, so very caucious to monitor them and pull earlier than they.
i do not see in text of announcment any reason to use copilot specifically anymore.
so left copilot 2 weeks ago.
tried local qwen-coder. started small but ended max fits 12 gb nvidia laptop gpu. it made my laptop sound like jet, so not this year or even next year, until bubble burst. no local for now. need pull earlier than other they from clouds so at some points.
but local was fine quality autocompletion even on 12gb.
currently ended up on vs code+continue+mistral. feels fine. i like line by line autocomplete and pressing tab several times for each line. better quality and more control.
also i use only clouds which also provide models to run locally. so i can pull when time will come.
do not have time to try other models and extensions. so many.
Why do I have the feeling my subscription to Github Copilot Pro since day 1 is going to feel a downgraded experience very soon?
The rugpulls shall continue until morale improves
I'll pass. I'm waiting for GitHub Copilot Pro+ Max.
And then for them to say that they have too many pricing tiers and consolidate all of them into GitHub Copilot One.
I bought a year sub to Copilot when they first announced it and I have no idea what I'm even getting from it anymore with all the changes lol
- https://github.com/cline/cline
- https://github.com/ggml-org/llama.vscode
- https://openrouter.ai
I'm sure there are others if you're interesting in switching to something else.
"Proudly announcing that GitHub Copilot is not unlimited anymore, enjoy!"
I only use Copilot for code completion, the other features are insufferably bad, at least in my IDE. I’m beginning to feel like I’m seriously overpaying. Does MS not update their completion model, or is it just my impression that suggestion quality has been stagnant for literally years? Do you guys know any cloud based code completion services with performance superior to Copilot?
Try Cody from source graph, its quiet light at least on VSCode + Mac and it is very customizable and has models across OpenAI and Anthropic
Does Cody have an agent mode? I have the free vesion and it doesn't seem to. IMO coding assistant without agent mode is useless, it's useful just only for asking questions.
Copilot used to be much worse than Cursor but they have agent mode and MCP now.
Also I use Cody as a backup because I thought their ollama support means I can code on the plane, but when I actually tried, it stills requiring SourceGraph access and failed.
I felt the same way in an earlier HN thread a few weeks ago and some suggested tools like Cursor.
Cursor has been way way better. And agent mode now makes me feel like there’s something more to this AI coding stuff than just fluff.
I also only use the code completion. I recently started using the Continue extension in VSCode running local models. They get the job done and it's free.
Supermaven is very good if you're using VSCode and not cursor/windsurf tab completes - both of which are significantly better than standard.
It is unfortunately also abandoned. The creator got an offer from Cursor to integrate it last Sepmtember and abandoned it, seems likely it was part of the contract. By the looks of it he also didn't refund subscribers, not a good look.
This is purposefully aimed at Cursor.
$10 from Github vs $20 for Cursor. [0]
And finally...
Extinguish. [0]
[0] https://github.com/features/copilot/plans
[1] https://github.com/getcursor/cursor/issues/2976
The licenses for the Microsoft extensions explicity don't let you use them in non-MS editors. They seem to be cracking down now.
To be fair, Copilot came before Cursor and the "agent mode" isn't a Cursor specific feature.
I use Cursor every day but I'm more than happy to switch if a better tool exists.
I'm personally wait for the JetBrains equivalent to cursor: https://www.jetbrains.com/junie/
I've had problems with pylance and auto complete on Cursor for months for (what I think is) the same reason - so I'm not sure the crackdown has only started in the last couple days
best move is to switch to based-pyright
i was in previous iterations of products ms doing eee/enshittification, so very caucious to monitor them and pull earlier than they.
i do not see in text of announcment any reason to use copilot specifically anymore.
so left copilot 2 weeks ago. tried local qwen-coder. started small but ended max fits 12 gb nvidia laptop gpu. it made my laptop sound like jet, so not this year or even next year, until bubble burst. no local for now. need pull earlier than other they from clouds so at some points. but local was fine quality autocompletion even on 12gb.
currently ended up on vs code+continue+mistral. feels fine. i like line by line autocomplete and pressing tab several times for each line. better quality and more control.
also i use only clouds which also provide models to run locally. so i can pull when time will come.
do not have time to try other models and extensions. so many.
God this is such a horribly written page. A great example of what not to do for product companies. I'll share this in my newsletter.