MakoBench results: Total 4.9 (suite 4.1)

#35
by MakoBench - opened

Scores for spark-x2.6-4b-q8-0 (suite 4.1):

Total: 4.9

Code: 2.7
Chat: 2.3

NSFW level: 1.9

Rank: 44 of 47 in suite 4.1
Quant tested: Q8_0
Tested: 2026-11-08

https://huggingface.co/spaces/MakoBench/makobench

SparkLLM org

@MakoBench Thanks for testing our model and sharing the results! We’ll be open-sourcing new models. Stay tuned!

This looks interesting, because any fresh new previously unknown test is unlikely to be contaminated by "benchmaxing".

For the coding tests, could you please also evaluate programming languages other than Python? I'm primarily interested in the good old C/C++ for my own projects. The other people may be interested in Rust, Golang or even Javascript.

Sign up or log in to comment