second_constantine/
magistral-small

371 3 months ago

Building upon Mistral Small 3.2 (2506), with added reasoning capabilities, undergoing SFT from Magistral Medium traces and RL on top, it's a small, efficient reasoning model with 24B parameters.(quantized UD-Q5_K_XL)

vision tools thinking 24b

Models

View all →

1 model

magistral-small:24b

18GB · 128K context window · Text, Image · 3 months ago

Readme

Feature Value
vision true (>=0.11.11)
thinking +/-?
tools true
Device Speed, token/s Context VRAM, gb Versions
RTX 3090 24gb ~42 4096 19 UD-Q5_K_XL, 0.12.2
RTX 3090 24gb ~42 15360 21 UD-Q5_K_XL, 0.12.2
M1 Max 32gb ~15 4096 18 UD-Q5_K_XL, 0.12.2
M1 Max 32gb ~15 15360 20 UD-Q5_K_XL, 0.12.2