Hacker News

Recursive Introduces Flint Model (recursive.com)

30 pointsby guybedo4 days ago6 comments

nvtop 3 days ago

- Better than Gemma 4 and Muse Glimmer on SWE benchmarks

- 125 tokens per second on an M4 Pro MacBook Pro

- 512k tokens context window

- 40B total, 2.5B active parameters mixture-of-experts

Pretty cool release!

dofm 3 days ago

I'm just about to test this one out (M1 Max).

I really like Muse Glimmer, and on low reasoning level it's both helpful and quite quick.

But I won't turn down a model that is faster.

Nemotron Bolt appears to be really quite good but unfortunately it is quite slow on macOS/Metal in llama.cpp at the moment because of lack of support for a tensor type (or something… not sure I understand).

kamranjon 4 days ago

Pretty amazing for being trained from scratch, to have ~Qwen 3.8 27b quality with such an increase in tok/s performance. Excited to try it out and see how snappy it is.

aistripes 3 days ago

Numbers seem good but can’t really be impressed on real world test https://hypeorcode.com/models/flint/

trehcrob 4 days ago

Congrats on the release, hope the next iteration outperforms the latest Qwen model.

nittanymount 2 days ago

not as good as qwen3.8 yet, even the chart says good numbers, but