Hey hey there. More genuinely grand videos. I'm really, really, I'll go so far as to say, "thrilled" with your direction and bluntness. I doubted the existence of someone making content that isn't simply regurgitating other's videos or word for word the write up on XYZ tool. I never imagined I'd be praising a TH-camr quite so much, but you're truly doing a great job making plenty of content while maintaining what seems like self-respect and morals. Congrats.
IBM Granite 3.0 was just added to Ollama. Could you please see how good this model is at coding. Their internal benchmark say it's better than mistral.
I know this is old, but I keep getting the error Invalid HTTP request received when trying to use with litellm. Installed the litellm[proxy] and still didn't resolve the issue. I'll keep searching
Is it possible to allocate the responsibilities of web app development among Aider, Cline, and Cursor according to their strengths to achieve the best results while minimizing costs?
Thanks a lot for inspiring videos, King! Could you please explore the possibility of integrating cline/aider with intelligent free tier model router? I mean we have a great pool of powerfult high-quota free models, such as Gemini Flash/Pro, all kind of Mistrals, portal versions of Claude and gpt4o - it would be really great to have an intelligent proxy that could route the prompt calls from cline/aider, considering the role, specifics and importance of each prompt - fundamental, difficult, architectural or debug prompts could go through Gemini Pro, Mistral Large or Claude/GPT4/Bing Copilot. That combined (and we can have several accounts) provides a good amount of free daily prompts - and the remaining routine coding could be done with Gemini Flash, Codestral and so on. Would be really thankful for such attempt!
what is the different between the codestral and mistral large ? which one is better for code and for daily chat ? if you prefer, which one is better between this model vs gpt plus or gemini pro 1.5 ?
Great video sir. I've been following along for a few months and am really liking my setup. Which do you think performs better in Cline, deepseek-chat or codestral?
@@AICodeKing Good to know. I dropped $5 on the API when you highlighted it a couple of weeks ago. Great value, it'll last me several months. I'll stick with that, thanks!
I actually did not followed the video well the first time, With help of Claude I was able to fix the issue, I realised after watching it again that you had already explained how to fix litellm and cline issue, Thanks Again King! (lol I've never called anyone King before, Been following you videos for a while and really appreciate all of your work!) - Just curious - Is it your real voice you use or AI lol - Also hooked into your video's intro music
With Aider, you can use o1 as the architect. For the coding, stay with cheaper models (Sonnet 3.5, Deep Seek, ...). With Cline, since you can't (for now) use 2 models at once, I recommand to stay with Claude Sonnet 3.5.
I tried Cline without any configuration, and then with gemini, it was pretty simple, even if I pointed to a complete example of another function. I then did with chatgpt o1 preview and it did the job.
By the way, I created an app 📱 inspired by my 3-year-old autistic son, Heitor 🧒🏻, to help with speech development. The app's name is SPEAK OUT KIDS. 🙏 The app is completely free, and it would mean a lot for me if you could test it, and if you find it relevant, share it to reach even more families with atypical children. 🙏♥️ It's also suitable for typical children, and has 9 languages to learn from 🇧🇷 🇺🇸 🇪🇸 🇩🇪 🇮🇹 🇫🇷 🇯🇵 🇨🇳 🇮🇳 🇰🇷 🇦🇪 🇷🇺
@@echofloripa o1-preview is really good... but very expensive. Claude Sonnet 3.5 will have approximately the same efficiency in coding for a tenth of the price and a lot faster. Gemini is (IMO) crap for coding.
I was happy with Aider, But then I switched to cline. I like the interface of cline better since you "Save" or "Reject" the things it does, which you cannot do with Aider.
Mammouth is good to replace your subscription to Chat-GPT, not OpenAI API, since they don't provide API. For using API in dev tools like Aider or Cline, OpenRouter is better.
hi, aicodeking! thank you so much for your honest hard work into learning this and sharing your knowledge. i followed your tutorial on installing mistral on cline, its all well, except when i run a task, it says: "500 No module named 'opentelemetry'". what could be wrong? thanks again!
@@salvolop9530 Did you find a method to use Nemotron in Cline? I surf the internet but can not figure it out. Please let me know if you succeed. Thx !!!
According to Artificial Analysis, Gemini 1.5 Flash is much better at coding (HumanEval) and the rest of params than Codestral. I think Aider benchmarks are completely outdated.
Codestral Mamba and Codestral are 2 different models. Mamba is a smaller 8b model while codestral is something like >20b. I just want to laugh at how easily people jump to conclusions after doing not even a little bit of research.
Hey hey there. More genuinely grand videos. I'm really, really, I'll go so far as to say, "thrilled" with your direction and bluntness. I doubted the existence of someone making content that isn't simply regurgitating other's videos or word for word the write up on XYZ tool. I never imagined I'd be praising a TH-camr quite so much, but you're truly doing a great job making plenty of content while maintaining what seems like self-respect and morals. Congrats.
Gemini is amazing! Very good for assistint NodeJS / Golang code.
Thanks for suggesting the Mistral one, I will try later.
Seems like we get a new best free way everyday
IBM Granite 3.0 was just added to Ollama. Could you please see how good this model is at coding. Their internal benchmark say it's better than mistral.
I know this is old, but I keep getting the error Invalid HTTP request received when trying to use with litellm. Installed the litellm[proxy] and still didn't resolve the issue. I'll keep searching
Can you explain about the terminal how to run, it litellm
did you get an answer?
@@IslamicVersion I don't know how to get it to run either @AIcodeking please help
good to see actually free as many times it's just the introductory trial before the costs kick in. but i'm using cursor, vO right now and loving it.
Is it possible to allocate the responsibilities of web app development among Aider, Cline, and Cursor according to their strengths to achieve the best results while minimizing costs?
API key has to look like and openAI key so make it something like sk-1234 once you have your LLM proxy set up.
On a roll!! Phenomenal
Very nice! Nice find on the litellm proxy.
Thanks a lot for inspiring videos, King! Could you please explore the possibility of integrating cline/aider with intelligent free tier model router? I mean we have a great pool of powerfult high-quota free models, such as Gemini Flash/Pro, all kind of Mistrals, portal versions of Claude and gpt4o - it would be really great to have an intelligent proxy that could route the prompt calls from cline/aider, considering the role, specifics and importance of each prompt - fundamental, difficult, architectural or debug prompts could go through Gemini Pro, Mistral Large or Claude/GPT4/Bing Copilot. That combined (and we can have several accounts) provides a good amount of free daily prompts - and the remaining routine coding could be done with Gemini Flash, Codestral and so on. Would be really thankful for such attempt!
Does mistral large have a free rate limit?
how about supermaven ? so how is your setup now ?
just curious aicoder what apps or websites have you made that you run to this day, any examples?
what is the different between the codestral and mistral large ? which one is better for code and for daily chat ?
if you prefer, which one is better between this model vs gpt plus or gemini pro 1.5 ?
Great video sir. I've been following along for a few months and am really liking my setup. Which do you think performs better in Cline, deepseek-chat or codestral?
If you pay then Deepseek otherwise for free, Codestral.
@@AICodeKing Good to know. I dropped $5 on the API when you highlighted it a couple of weeks ago. Great value, it'll last me several months. I'll stick with that, thanks!
Is it possible to utilize multiple AI models within a single project using this?
Thank you, this video is very helpful 👍🏾
Aider and cline does the same thing. What's the difference?
Cline has UI
@@PARADOX885 is this the only difference or their performance also vary?
it's the API part that they both use instead of claude 3.5 sonnet, gemini as he mentioned
yes
@@tradingwithwill7214 We can set gemini or any available models in both aider and cline. Am i right?
how to setup litellm? followed every step as per video but getting API error in Cline
I actually did not followed the video well the first time, With help of Claude I was able to fix the issue, I realised after watching it again that you had already explained how to fix litellm and cline issue, Thanks Again King! (lol I've never called anyone King before, Been following you videos for a while and really appreciate all of your work!)
- Just curious - Is it your real voice you use or AI lol
- Also hooked into your video's intro music
@@Dangerkabaap He is using Runwayml TTS voice.
thanks, you are the king!!! ;)
would OpenRouter work instead of LiteLLM?
No, it would cost you money.
I'm a non-programmer and am intimidated by Aider😬
Can I somehow use the Mistral models in VSCode+Cline without having to install/use Aider?
yes you can
@ecoBag-777 thanks, could you maybe point me to how this is done?
@@DavidG2P i will create a video just for you my guy
@@ecoBag-777 4-5 search terms would probably suffice
It's worth to use gpt4 o1 preview or or o1 for coding with Cline?
For flutter, anyone knows which is the best model?
Generally, Claude is the best.
With Aider, you can use o1 as the architect. For the coding, stay with cheaper models (Sonnet 3.5, Deep Seek, ...).
With Cline, since you can't (for now) use 2 models at once, I recommand to stay with Claude Sonnet 3.5.
I tried Cline without any configuration, and then with gemini, it was pretty simple, even if I pointed to a complete example of another function. I then did with chatgpt o1 preview and it did the job.
By the way, I created an app 📱 inspired by my 3-year-old autistic son, Heitor 🧒🏻, to help with speech development. The app's name is SPEAK OUT KIDS. 🙏 The app is completely free, and it would mean a lot for me if you could test it, and if you find it relevant, share it to reach even more families with atypical children. 🙏♥️ It's also suitable for typical children, and has 9 languages to learn from 🇧🇷 🇺🇸 🇪🇸 🇩🇪 🇮🇹 🇫🇷 🇯🇵 🇨🇳 🇮🇳 🇰🇷 🇦🇪 🇷🇺
@@echofloripa o1-preview is really good... but very expensive.
Claude Sonnet 3.5 will have approximately the same efficiency in coding for a tenth of the price and a lot faster.
Gemini is (IMO) crap for coding.
don't yout get llama 3.1 405B for free with openrouter too? i get that mistral large is good, but the others?
If you have issues, it will likely be due to your python version, pip version.. I ended up going back to 3.10 and -m pip instead of pip3??
Yes, Python 3.13 is not supported by some tools yet. Even 3.12 support added just recently. 3.10 is safe
Any video on IBM Granite coming soon??? I want to know if it can code at all
@@emmanuelkolawole6720
For now, it cannot use it in Cline at all.
I've dowload 4 different size of 3.0 models and none works.
@@shihfamily4301 Can it use it in Aider?
Aider is the best ai agent❤❤
I was happy with Aider, But then I switched to cline. I like the interface of cline better since you "Save" or "Reject" the things it does, which you cannot do with Aider.
@@WaytogoforAI aider has an undo function which is similar
Oh mann cline is a life saver.
@@Valk-Kilmer What do you mean? 🤨🤣
CodeChadKing, what are your thoughts on Mammouth AI? is it a better option?
I haven't tried it.
Mammouth is good to replace your subscription to Chat-GPT, not OpenAI API, since they don't provide API.
For using API in dev tools like Aider or Cline, OpenRouter is better.
First... it's pretty cool...
I dont know how to run litellm commands
Just like you run aider commands.
@@AICodeKing yeah, I get it, thanks. But it is like the litellm is not a internal command... I will try to figure it out
@@AICodeKing Tested with success, thanks :)
hi, aicodeking! thank you so much for your honest hard work into learning this and sharing your knowledge. i followed your tutorial on installing mistral on cline, its all well, except when i run a task, it says: "500 No module named 'opentelemetry'". what could be wrong? thanks again!
Are you on windows?
i guess you need to pip3 install opentelemetry but that's weird
qwen 2.5 72b is also almost as good as claude sonnet 3.5 in coding
Can it run locally on a single Nvidia 3090 GPU ? Or is there any free or cheaper provider choice to use the API for Cline ?
Compared to Nemotron?
I doubt that
@@salvolop9530 Did you find a method to use Nemotron in Cline? I surf the internet but can not figure it out. Please let me know if you succeed. Thx !!!
@@mik3lang3lome neither. I've tried use qwen and sonet side-by-side to do the same coding task to compare, and IMO qwen is still far away from sonet.
thank you ❤
How do I run those pip installs on windows terminal
It's same. Just install python and it should run.
@@AICodeKing thank you. I'll get back with results.
Make sure you install pip/pip3.
which olama model is best for code generation using cline for free
Qwen-2.5 coder
@@AICodeKingis deepseek is better than qwen ?
only one Qwen-2.5 coder tool works for me
But it seems quite slow
is it completely free api or free for trial period?
Completely free
it's obvious that those 'free' models are cashing in on your data for their training
I've been using deepseek coder... will i shift... hm....
I mean it's free..
is deepseek best for code generation? is it free?
@@pomodoro_work for me yes, compared to soonet no... sad eh...
how about you guys?
According to Artificial Analysis, Gemini 1.5 Flash is much better at coding (HumanEval) and the rest of params than Codestral. I think Aider benchmarks are completely outdated.
Codestral Mamba and Codestral are 2 different models. Mamba is a smaller 8b model while codestral is something like >20b. I just want to laugh at how easily people jump to conclusions after doing not even a little bit of research.
I didn't realized that I am not first 😅
Dang never know litellm exists, very useful
First!