Qwen3 VL support merged into llama.cpp

Syndicated from the fediverse. Read and engage on the original instance.

Benchmarks look pretty good, even better than some of the text only models, make sure to take them with a grain of salt tho

::: spoiler Qwen3 VL 30b a3b (No Thinking) :::

::: spoiler Visual benchmarks for Qwen3 VL 235 A22B (Thinking) :::

3 replies

i am usng it with openwebui and wireguard at school i just upload a pic of my paper and it does all the problems for me

4/6 bait

devxyn reply

I’ve been trying the 32b instruct variant at Q4_K_M and it’s been solid for general use, tool use, and image comprehension. Pretty impressive

-1