Reflection Open-Sources New Model Amid Training Efficiency Debate

Reflection released a new open-source model under Apache 2.0, but analyst estimates of 12% pre-training MFU sparked criticism. The company responded that actual MFU is slightly above 19%, with goodput reaching 92% in under four weeks of training.

2026-10-06 ~ 2026-10-06 · 2 related posts

• Reflection says Beam hit ~19% MFU with 92% goodput during its 4-week RL run — alexpolozov · 2026-10-06
• Reflection Ships Apache 2.0 Model With Tech Report; Analyst Estimates Pre-training MFU at Just ~12% — eliebakouch · 2026-10-06