Vibeleaderboard
Sao10k

model publisher

Sao10k

This is a Model Atlas entity, not a ranked standalone lab. A community or pseudonymous model publisher, not a verified standalone lab.

Sao10K is a notable example of individual fine-tuners creating persistent model families for preference domains that general leaderboards rarely measure well.

Models in the Atlas →

Profile

Overview

Sao10K is a pseudonymous open-model creator focused on creative writing and roleplay fine-tunes. The Euryale and Lunaris releases adapt Meta Llama checkpoints for character interaction, prose, and less constrained narrative behavior.

Weights and model cards are distributed through Hugging Face, with hosted access available from model routers. There is no conventional corporate profile or published research program behind the identity.

Notable contributions

  1. 01Long-running roleplay fine-tune lineageEuryale carried a recognizable creative-writing objective across multiple generations and parameter scales of Llama.