Today we're releasing Underdog Saluki 27B
Based on Qwen 3.8 27B, Saluki is almost 7x smaller than its full-precision counterpart while it keeps 96% of its benchmark performance and beats Qwen at tool calling
It's the best model <8GB that runs in Underdog
Saluki is available today under Apache 2.0. Try today in Underdog https://t.co/wAgfW3X3ok
Today we're releasing Underdog Saluki 27B
Based on Qwen 3.8 27B, Saluki is almost 7x smaller than its full-precision counterpart while it keeps 96% of its benchmark performance and beats Qwen at tool calling
It's the best model <8GB that runs in Underdog
Saluki is available today under Apache 2.0. Try today in Underdog https://t.co/wAgfW3X3ok
great , A 2-bit 27B just beat Qwen 3.8 on tool calling.
Underdog Saluki shrinks the model from 54GB to 7.89GB and Overall retention sits at 96%. scores higher on their tool-call bench (and parallel calls) under the same harness.
A full-size agent now fits on a 12GB card with room left for context.
That’s the part agents actually run on.
Math and reasoning drop more check the card before you swap your daily driver.
Try Underdog: free, capable, private AI running locally on your device
New build and Saluki 27B are out. Under 8GB, beats Qwen 3.8 27B at tool calling
Like, repost, and comment one task you wouldn’t give a cloud agent to get an invite🦴
Today we're releasing Underdog Saluki 27B
Based on Qwen 3.8 27B, Saluki is almost 7x smaller than its full-precision counterpart while it keeps 96% of its benchmark performance and beats Qwen at tool calling
It's the best model <8GB that runs in Underdog
Saluki is available today under Apache 2.0. Try today in Underdog https://t.co/wAgfW3X3ok