MMGA / COMMUNITY MODEL LABModel Leaderboard
BSC NETWORK
COMMUNITY MODEL LAB

Back the next
MiniMax H3 fine-tune.

Discover community-trained H3 models and use staked MMGA to decide which models the network should support next.

MODEL PROJECTS4 H3

Community fine-tunes

TOTAL STAKED VOTES4.92MMMGA

+8.2% this epoch

EVALUATION LEADER96.8score

H3 CinemaFlow

30D MODEL JOBS182Kjobs

Across all projects

MINIMAX H3 COMMUNITY FINE-TUNES

Model Leaderboard

Rankings combine community support and evaluation score. Every vote consumes staked-MMGA voting power.

RANK / MODELSPECIALTYEVALUATIONSTAKED VOTES
#1
H3 CinemaFlow@motion-labs · Community fine-tune
Cinematic camera movementMiniMax H3 base
96.8
1,842,000staked MMGA
#2
H3 ProductCraft@commerce-guild · Community fine-tune
Product and commercial videoMiniMax H3 base
94.1
1,368,000staked MMGA
#3
H3 AnimeMotion@cel-community · Community fine-tune
Anime motion and consistencyMiniMax H3 base
91.7
987,400staked MMGA
#4
H3 WorldSim@open-worlds · Community fine-tune
Physical motion and environmentsMiniMax H3 base
89.4
721,600staked MMGA
Stake-weighted model curation

Staking MMGA creates voting power; community votes signal which fine-tunes should enter technical evaluation and potential network support. A vote is not an automatic deployment.

Get voting power