It did not perform as well as I expected. Qwen2.5-Coder-3B (2 years old) outperformed it by a wide range -> fixing ~50% of bugs whereas this model only fixed ~12%.
Granted, it's not a coder specific model, but given its benchmark performance to Gemma models, and that it's two years newer, and that it's an MoE with 8B total params, I expected it to be more competitive.
I would much rather not run the model on my local laptop hardware and offload that to some system sitting under my desk in my home office, accessible via VPN, than take the risk of using an unreliable and flaky tool for the convenience of having it on the same hardware on my lap.
I pay very little attention to 8 billion or whatever (or even much smaller) models these days and I don't feel like I'm missing much.
LLamaCPP has had some massive updates in the last week or so.
Using Vulkan backend.
``` llama-server -fa on -t 7 -ngl 999 --mlock --fit off --kv-offload --no-webui --metrics --chat-template-kwargs {"preserve_thinking": true} -b 2048 -ub 1024 -m /mnt/models/unsloth/Qwen3.6-27B-MTP-GGUF/Qwen3.6-27B-UD-Q5_K_XL.gguf --mmproj /mnt/models/unsloth/Qwen3.6-27B-MTP-GGUF/mmproj-F16.gguf -c 262144 --kv-unified -ctk q8_0 -ctv q4_0 --spec-type draft-mtp --spec-draft-n-max 3 --spec-draft-ngl 99 --alias unsloth/Qwen3.6-27B-MTP-GGUF --temp 0.60 --top-k 20 --top-p 0.95 --min-p 0.00 --presence-penalty 0.00 --repeat-penalty 1.00 ```
I think some of the folks in the local llm social media communities are using them for things like company-hosted customer service chat bots, or purely english text writing stuff where Q4 will probably not cause a problem. For more discrete technical work I stick pretty much exclusively to Q8.
Smaller quants for the model start to fall off but more importantly, smaller KV-cache quants fall off much faster so avoid less than Q8 there.
Edit: Is it this? https://github.com/Liquid4All/cookbook/tree/main/examples/lo...
FYI: Opencode is very well tuned for Qwen models, but I haven’t found it that rare for niche models to perform badly in it.
If I were going to use this model, I'd be looking to use it more as is the primary chat interface of a larger system, and having it orchestrate & delegate tasks to other places via tool calls. It's not quite as exciting on the surface as a local "do it all" model, but it does enable some pretty neat use-cases, IMO.
I'm imagining a local agent that is super low latency, works entirely offline, and capable of queuing up complex tasks for larger/smarter cloud agents which execute them asynchronously.
Two of the other responses speak about it being abysmal at tool calling.
Overall, I'm pretty impressed a model this small can find/fix ~12% of bugs with crappy context - even if they're about as easy as possible to fix.
I just assumed it would perform better, given all the advancements in the space.
It's possible 1B active parameters is just not enough - even if it has 8B params of knowledge to reason through bugs.
Playing around with the context I fed it, it was able to fix up to ~34% of bugs vs ~46% for Qwen2.5-Coder-3B and ~54% for Qwen2.5-Coder-7B.
But even against E4B it's shaky, which is surprising given how many tokens they trained on. I guess it was on a lot of synthetic data.