So I don't really talk about this often and I haven't cared about LLM's or local LLM's until just recently. I have to say, Qwen 3.8 27B is wild... If you don't know what I'm talking about, I'm talking about a large language model consumers can run on their own computers. Of course, you still have to pay over $1,000 on a GPU, Mac, or DGX Spark in order to have good speeds and high quality outputs but the point is that you can download the model and run it.
But the thing that's blowing me away is how good it is. I recently tried out Claude Sonnet 5 to create some command line scripts and I was surprised when it one-shot one of my tests. At that moment, I realized what the whole "agentic coding" thing was all about and why "agentic AI" is a big deal.
Then I heard that Qwen 3.8 27B was releasing and many expected it to rival Claude Sonnet 5 in some way. Such a claim as quite optimistic... But after playing around with it myself and seeing the benchmarks and the scoring from Artificial Analysis, it's the real deal. Before the release of 3.8 27B, I was trying to use Qwen 3.6 27B and it was generally a disappointment in regards to chat and programming. It couldn't one-shot some of the command line app tests I put it through. But this new model from Qwen is and I think it's better for conversation as well.
Lemme just put it like this... For those with the hardware, we now have something that can create us lightweight programs on a whim. This is huge because now anyone anywhere can have access to more powerful, nimble, focused software around the world. You can even get the LLM to search the internet for you and it can often provide reasonable summaries, evaluations, or explanations based on your questions. It can even help with creative writing. When people were saying this LLM AI tech was a big deal, I'm starting to understand why. It's not just for entertaining ourselves, it can be a real powerhouse when it comes to work and boosting one's own productivity.
For a long time now I haven't bothered to use or care about LLM's. But now that I have one that lives on my own computer that's around the quality and depth of the frontier proprietary options, I'm definitely appreciating the tech now. I think this particular Qwen LLM is absolutely revolutionary for humanity. Hopefully as time goes on, this level of intelligence/power is made more efficient and shrunk down to a size of a phone or raspberry pi. Even better is if what I'm running on right now could run on an old CPU at blazing speeds. I'm quite optimistic and it also goes to show that the "quality floor" has risen all around, especially here on the internet.
But the thing that's blowing me away is how good it is. I recently tried out Claude Sonnet 5 to create some command line scripts and I was surprised when it one-shot one of my tests. At that moment, I realized what the whole "agentic coding" thing was all about and why "agentic AI" is a big deal.
Then I heard that Qwen 3.8 27B was releasing and many expected it to rival Claude Sonnet 5 in some way. Such a claim as quite optimistic... But after playing around with it myself and seeing the benchmarks and the scoring from Artificial Analysis, it's the real deal. Before the release of 3.8 27B, I was trying to use Qwen 3.6 27B and it was generally a disappointment in regards to chat and programming. It couldn't one-shot some of the command line app tests I put it through. But this new model from Qwen is and I think it's better for conversation as well.
Lemme just put it like this... For those with the hardware, we now have something that can create us lightweight programs on a whim. This is huge because now anyone anywhere can have access to more powerful, nimble, focused software around the world. You can even get the LLM to search the internet for you and it can often provide reasonable summaries, evaluations, or explanations based on your questions. It can even help with creative writing. When people were saying this LLM AI tech was a big deal, I'm starting to understand why. It's not just for entertaining ourselves, it can be a real powerhouse when it comes to work and boosting one's own productivity.
For a long time now I haven't bothered to use or care about LLM's. But now that I have one that lives on my own computer that's around the quality and depth of the frontier proprietary options, I'm definitely appreciating the tech now. I think this particular Qwen LLM is absolutely revolutionary for humanity. Hopefully as time goes on, this level of intelligence/power is made more efficient and shrunk down to a size of a phone or raspberry pi. Even better is if what I'm running on right now could run on an old CPU at blazing speeds. I'm quite optimistic and it also goes to show that the "quality floor" has risen all around, especially here on the internet.
2❤️1
