Ministral (Fully Tested) : This NEW Mistral Model is the Llama-3.1 REPLACEMENT! (Good at Coding!)

Join this channel to get access to perks:

In this video, I'll be telling you about the new Ministral (3B & 8B) Models by Mistral AI and we'll be fully testing them as well. This is a new fully Opensource Model that we can host locally as well. It is also claimed to be better in Coding Tasks and is also really good at doing Text-To-Application, Text-To-Frontend and other things as well. I'll be testing it to find out if it can really beat other LLMs and i'll also be telling you that how you can use it. This models beats Qwen-2, Llama-3.2, Mistral, GPT-4O-Mini and Others!

----
Key Takeaways:

📈 Major AI Model Releases: The video covers two exciting AI releases, Nvidia Nemotron and Mistral Ministral, comparing them to Claude 3.5 and Llama-3.1 models. Stay updated on the latest in LLM advancements!

⚡ Ministral's New AI Models: Discover the 3B and 8B Ministral models, which feature function calling and a huge 128K context length, pushing boundaries in AI model performance.

💡 Benchmark Controversy: Dive into the benchmarking issues of Mistral’s models and why their "Knowledge and Commonsense" benchmark might be a little misleading. AI benchmarks are crucial, but manipulation is common.

🚀 Open-Source AI: While the 8B model is open-source, the 3B model is restricted to API-only access, leaving developers with limited options for experimentation. Learn more about the Mistral Research License.

💻 Coding Test Results: This video tests the 8B model against various coding challenges, such as building a confetti button and Python scripts, revealing its chain-of-thought reasoning and effectiveness in solving problems.

📊 AI Model Performance Comparison: A detailed comparison between Ministral, Llama-3.1, and Qwen-2.5, showing how Ministral holds its own in certain areas but falls short in cultural and general knowledge.

🎥 Upcoming AI Content: Stay tuned for a deep dive into the Nvidia Nemotron model, which promises to be a game-changer in the AI space. Don't miss out on these valuable insights!

-----
Timestamps:

00:00 - Introduction
00:46 - About Ministral
03:09 - Testing
07:57 - Final Results & Charts
09:20 - Ending

[Luceo-xrhb]

The model itself is really cool, but the license is a big minus.

[mal-avcisi]

We need a 1b model which csn beat o1 preview

[vaioslaschos]

I kind of agree that these extra benchmarks are very often introduced for manipulation, however we need more new benchmarks. And is not that there is some authority on benchmarks. For each of them, some student, some department, some company came up with. It is very hard to capture the experience of using an LLM with benchmarks though. gpt4o appears to be best in everything at the moment, but I still trust claude most of the time.

[jackfrost]

The release of Claude 3.5 Opus is rumored to be scheduled for the 22nd of next week by jimmy apples

[cbgaming]

We need a Nemotron 70B model video from you, sir @AICodeKing

[abdullahahmed]

In your next video, please also share sources where we could find free APIs for Nemotron and Ministral

[vladimirbatalin]

@AICodeKing, I still cant get this silly "Whai is C doing?"question.
In my world C he still kicking E's ass at the table.
Is there something i'm missing in a table tennis game rules in 2024?

[fkuz]

ok so today I gave a try to Nemotron 70b and It almost kill my PC :( it's a performance beast.
I have RTX 3060 TUF GAMING V2 12GB
i9 14900F
64GB RAM

What free LLM is best for coding in your opinion - I do not want to pay for tokens. Is there any which is comparable to Claude 3.5 / GPT4o?
Thank you! great work bro

[zeusconquers]

code king please do a review of aria moe by rhymes ai. that one has caught my attention. im just going to run it and see whats up

[ppbroAI]

You focus a lot on the coding side of LLM´s, but I think there is not a video deciding the best coding model for consumer hardware. I think qwencoder, but Deepseekcoder Lite, Codestral? I don´t know!, . We need a video from you 🤔

[abdullahzafar]

I wonder if the nemotron model will work on my All AMD Laptop ...

I use Llama 3.1 70b works fine on cpu, because mu GPU only has 12GB Vram

[abrahamsimonramirez]

isn't mistral-nemo 12B much better?

[HomeDev]

@AICodeKing
It is correct about the hexagon's diagonal, because (128 x sqrt(3))/3 = 73.9

[ChronicleContent]

what was the chat interface where you did the tests?