▲ 4 signals · 2 sources
[AINews] Inkling 2 Small: 17A-594B trained in 68 hours on 12T tokens, beating Muse Spark
Inkling 2 Small, a 594B-parameter mixture-of-experts model with 17B active parameters, was trained in 68 hours on 12 trillion tokens and scores higher than Muse Spark on the Artificial Analysis Intelligence Index. It supports image inputs and variable effort.
models latent.space ·
