List your company

Generative Recommendation on MT

11.7Latency (ms)MTServe
8.87227.96147.0566.139Apr 24, 2026

Evaluation results

MethodLinks
MTServeBatch Size=12026.0411.71.2160.2696.73
GPU-OnlyBatch Size=12026.0412.81.160.2660.26
Recomp.Batch Size=12026.0414.11——
MTServeBatch Size=42026.0417.32.5260.6997.44
MTServeBatch Size=82026.0426.63.161.7398.71
GPU-OnlyBatch Size=42026.0428.91.5160.6960.69
Recomp.Batch Size=42026.0443.61——
GPU-OnlyBatch Size=82026.0448.61.761.7361.73
Recomp.Batch Size=82026.0482.41——