Marin 535B open-source model training starts with 2.7e24 FLOPs plan
DavidBennett__ · x · 2026-08-24
Percy Liang announced the start of training for the Marin 535B-A23B model. The project is fully open, including code, data, recipes, and experimental results. Training will run on 11 GB200 NVL72 clusters for approximately 3 months, involving a total compute of 2.7e24 FLOPs across 18.75T tokens for pre-training and mid-training. The team previously debugged and forecasted the run using a scaling ladder from 1.6B to 27.7B models.
More from Infra
- NVIDIA and Sweden's Wallenberg launch Sferical One, a shared AI supercomputer — nvidia · 2026-10-16
- Next infra frontier: self-evolving systems that adapt and self-verify — _shreya_s · 2026-10-16
- Xiaomi's MiMo RL dashboard reveals $3.47M spent on just 30 training steps — bycloud · 2026-10-16
- Cerebras pushes disaggregated inference: heterogeneous prefill pools feeding wafer-scale decode — ai · 2026-10-16
- d-Matrix GA's Corsair rackscale, demos first public Attention-FFN disaggregation at 845 tok/s — ai · 2026-10-16
- Actual's inference engine hits 10k tok/s on NVIDIA DGX Station, GB300 case study — natesiggard · 2026-10-16