Review in real world tasks?

#19
by DuongLeVan - opened

This model shows strong benchmarks in coding and some others. Even beat Qwen3.8 27B in coding.
But have you guys tested it in your tasks?
Can anyone share some evaluations? Especially in coding tasks.

In another thread, they reported that it tries to output too much at once, and I agree. Otherwise, it's real. It's legit. It's a very solid model.

https://msnab.pages.dev/compare/#a=thinkingmachines-inkling-2-mini&b=unsloth-qwen3-8-27b-nvfp4

Sign up or log in to comment