this post was submitted on 23 Feb 2026
683 points (97.6% liked)
Technology
82414 readers
2961 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 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Well it is a 9B model after all. Self hosted models become a minimum "intelligent" at 16B parameters. For context the models ran in Google servers are close to 300B parameters models
Any source for that info? Seems important to know and assert the quality, no?
Here:
https://www.sitepoint.com/local-llms-complete-guide/
https://www.hardware-corner.net/running-llms-locally-introduction/
https://travis.media/blog/ai-model-parameters-explained/
https://claude.ai/public/artifacts/0ecdfb83-807b-4481-8456-8605d48a356c
https://labelyourdata.com/articles/llm-fine-tuning/llm-model-size
https://medium.com/@prashantramnyc/understanding-parameters-context-size-tokens-temperature-shots-cot-prompts-gsm8k-mmlu-4bafa9566652
To find them it only required a web search using the query local llm parameters and number of params of cloud models on DuckDuckGo.
Edit: formatting
Appreciated. Very much appreciated!
Not sure how we're quantifying intelligence here. Benchmarks?
Qwen3-4B 2507 Instruct (4B) outperforms GPT-4.1 nano (7B) on all stated benchmarks. It outperforms GPT-4.1 mini (~27B according to scuttlebutt) on mathematical and logical reasoning benchmarks, but loses (barely) on instruction-following and knowledge benchmarks. It outperforms GPT-4o (~200B) on a few specific domains (math, creative writing), but loses overall (because of course it would). The abliterated cooks of it are stronger yet in a few specific areas too.
https://huggingface.co/unsloth/Qwen3-4B-Instruct-2507-GGUF
https://huggingface.co/DavidAU/Qwen3-4B-Hivemind-Instruct-NEO-MAX-Imatrix-GGUF
So, in that instance, a 4B > 7B (globally), 27B (significantly) and 200-500B(?) situationally. I'm pretty sure there are other SLMs that achieve this too, now (IBM Granite series, Nanbiege, Nemotron etc)
It sort of wild to think that 2024 SOTA is ~ 'strong' 4-12B these days.
I think (believe) that we're sort of getting to the point where the next step forward is going to be "densification" and/or architecture shift (maybe M$ can finally pull their finger out and release the promised 1.58 bit next step architectures).
ICBW / IANAE