this post was submitted on 17 Jun 2026
165 points (97.1% liked)
Technology
85494 readers
4362 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 3 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
The vast majority of users ain't running anything but 27b max, more likely 14b, and that shit just ain't nearly as good as older saas models much less dominant like opus. Maybe for small shit but complex talks just ain't fitting on home hardware.
Completely agree, I forgot to mention that part. I am testing a few models ranging from 18b to 26b on my 7900xt. It is far from "make this complete system", but it can handle some smaller tasks. I think that will be the end goal anyway since cloud models fail a lot at maintainability, security, and other higher levels of thought that goes into coding. They can make a convincing prototype but I wouldn't hook it up to production.
Local models are already functioning well as a force multiplier. It can help explain logic, do minor refactoring, debugging etc. but with a bit of latency. I do think this is where we're headed since the frontier models required for generating a full prototype can't make production quality code and it is prohibitively expensive to do so. As far as I've heard, they're generally running spending ten times as much as they earn per token.
My guess is the next big thing to come out is, we can probably squeeze a lot more reliability out of smaller models. But their workflows, context, validations, etc will need to be very tightly optimized.
I can see harnesses coming with their own highly specialized lightweight models in the future. Some for very efficiently converting a basic prompt into chain-of-thought steps. Some for very efficiently determining relevant parts of a repository. Some for… a lot of highly specialized stuff. Then the harness would orchestrate these under the hood, reducing the cognitive load placed on any larger generalized LLMs. Those “larger generalized LLMs” could be something like 12b parameters.
Hopefully, soon after, we can start benchmarking how much different harnesses and augmentations improve baseline model performance. Ideally, in the long run, with a deeper understanding of how to tailor harness to workload and produce more procedural determinism. Then we can start configuring harnesses like data pipelines and run them through higher-level orchestration like Airflow too.
Sadly, that's true
Tried to refactor a spaghetti code state machine and thought, well, AI should handle this well. All the logic is there, just separate it into small functions to clean up the large one.
None was able to, alone because of the context window already
To be fair though, I tried Mistral online and it also stumbled around. ChatGPT was a complete clusterfuck - haven't tried Claude.
To be even fairer... it's a really large state machine, which was written on site during a fever and in stress - so... To defend myself a bit as well, how it even came to that ;-)
But seems, I'll need to go through this myself
Actually thought, that this would be a perfect example for using AI...
Yeah LLM's can help with many tasks but then there are times they just spout nonsense, or syntactically correct nonsense, the model size and context window just changes when they hit their limit.
Sometimes you have to call it quits, and try another way.