Which ChatGPT alternatives are you using?

[JosephTLyons]

For those of you using something outside of ChatGPT, what are you using and why? Please search for and upvote existing replies.

[widersky]

JetBrains LLM (in closed-beta) is unbeatable for me in case of accuracy and actual help in even more complex problems. CoPilot gets lost to a bit more complicated issues (for example, it can reverse the logic of the code), and ChatGPT itself is even worse 🤷‍♂️

[BKG01]

Any way to sign up for access?

[TerminalFi]

Honestly, this should live in a plugin. Just sounds like feature bloat

[enzime4u]

Using Codeium , it amazing and also FREE ! Does an amazing job

[gldtn]

Maybe Google Bard is a good option here, although it's still behind chatGPT. Also I believe their API is still in beta and you have to get on a waiting list.

[drgarlic]

Any from this list: https://ollama.ai/library, just being compatible with ollama's server would be awesome

[YurySolovyov]

Which in turn uses/wraps llama.cpp server, but both are great projects ✨

[chuikoffru]

It would be great to use codellama (ollama) as AI assist instead the Copilot in Zed.

[damywise]

Idk the context but I assume you mean in general?

1. Phind (vscode extension and web) for mostly programming, on vscode I use it along with Codeium.
   Codeium has auto complete but phind has better capability.
2. Perplexity for more broad general search. If I can't find something with Phind, sometimes I can find it in Perplexity, and vice versa.

I use them interchangibly.

[LucasJohnNyamhanga]

Am on GitHub Copilot.

[Pipboyguy]

Mistral Medium is fantastic

[RoninReilly]

https://openrouter.ai

[rohanpoudel2]

I am using Phind

[mindaslab]

https://tabby.tabbyml.com/ I am using it in VSCodium. This is also available as language server, so may be it might be easy to integrate into Zed.

[SMak06]

Can I use any of the alternative in Zed instead of chatGPT?

If so, how do I remove the existing API key that I entered (it's not working for me)?

[Moshyfawn]

You have already asked how to reset the API key three times, and one of those threads already contains the answer to your other question. Please refrain from posting the same question multiple times in different threads.

[OscarJPrice]

You can use Github copilot, but that appears to be it.

[simonmcconnell]

Cody by Sourcegraph, because it is the first and only one I've tried.

[tumma72]

I agree Cody in VS Code is at the moment the superior choice. The ability to switch model and the ability to keep context of the local code in a vectorized and embedded format make it an extremely valuable asset. Additionally the freedom to switch model and even use local Ollama installed modem when offline is great.

[joshrutkowski]

CodeWhisperer

[Moshyfawn]

Have you been using CW since Fig or just CW?

[joshrutkowski]

I’ll clarify CW (as in rebranded Fig), yes I use that. But primarily I meant the IntelliJ and VSCode extensions.

[KoStard]

+1 of codewhisperer, would be happy to see support for it in Zed

[Frank-vdm]

Im using a combination of copilot chat, openIA and a few others, in usually it depends on my project, ideally i would like to be able to configure which assistant to use in the assistant panel.

that being said since we already have github copilot intergration could we expect github copilot chat to become available?

[AugustinMauroy]

I'm using mistral
it's open source

[phtn]

please add phind

[clearfram3]

I've used Cody from SourceGraph, Double.bot, and GPT4 in chatGPT. Mistral's Medium, 8x7, and Large are really good, and Claude 3 looks amazing.

[jackbravo]

+1 to double.bot and Claude 3

[ggoonnzzaallo]

I'm one of the co-founders of double.bot, appreciate the shoutout! If there's anything we can do to improve the product, let me know :)

[tendiesfortwo]

+1 for double.bot!

[adamierymenko]

I've had decent results from DeepSeek Coder and CodeLlama models running in lm-studio locally using plugins in VSCode like Continue. LM-Studio is a local GUI for running local models that can start its own local OpenAI-compatible API server. (Bonus: running large local LLMs is one of the only things I can do on my M1 Max to make the CPU fans spin like an old Intel MacBook 😺 )

I'd really like to see support for configuring your own AI endpoints and using things like lm-studio and ollama.

Claude 3 is also excellent and if/when it's possible to integrate I'd love to see support for that too.

Is there a way to sponsor features?

[tendiesfortwo]

First step is ditching the OpenAI models (at least until they release something better). Claude 3 Opus is clearly superior.

Within the tools available right now that use Claude 3 (not many tbh), my favorite so far has been double.bot . Super generous free trial too to get a feel for it

[logiota]

Ollama ftw!

[phaynes]

In addition to chatGPT, I use Gemini and Claude. I think the Groq is going to be a transformational technology.

As I am a software architect, I do more text editing in markdown than most. The different engines allow me to review documents t to a higher standard than another by it self.

[enzime4u]

which ones are supported by Zed ?
Also is there a new guide to setup ChatGPT on Zed ? since it's a pain to set it up

[jenil777007]

ChatGPT & Claude 💯

[tokiory]

Definitely Codeium.

[ggoonnzzaallo]

Anyone using VS Code as their main IDE should check out Double.bot