Open Athena Starts Training Marin: 535B-Parameter MoE on 18T Tokens, Fully Open
TheTuringPost · x · 2026-09-03
Open research hits a milestone: Open Athena has kicked off training on Marin, a 535B-parameter (23B active) MoE model across 18T tokens, now 13% through the run.
- Every code pipeline, training log, and checkpoint is fully open for researchers to inspect
- Compute funding comes from the Jen-Hsun and Lori Huang Foundation via CoreWeave, channeling grants straight to academic labs at Stanford, MIT, Cornell, Princeton, and NYU
- Led by Percy Liang, David Hall, and team
The story matters for preserving academic agency: without such platforms, non-commercial researchers are priced out of frontier exploration entirely.
Related event: Open Athena Launches Marin, a 535B-Parameter MoE Trained in the Open(2 posts)→
More from Models
- GPT-6 Astra and Astra Aeon spotted in Codex; Aeon tipped as long-horizon agent model — VraserX · 2026-09-03
- Ant Group Open-Sources Finance-Enhanced Model Ling-3.0-flash-Fin with 124B Parameters — niacolhealth · 2026-09-03
- Should 'hard scientific problems solved' be the new LLM benchmark? — Dr_Singularity · 2026-09-03
- Ben Thompson on Gemini: 'OK, it's now or never' — kieranklaassen · 2026-09-03
- User Reports ChatGPT Randomly Outputs 'GOD IS COMING' and Won't Stop — jaypro1005 · 2026-09-03
- Ling-3.0-flash-Fin Weights Released: 124B Params, 5.1B Active, 256K Context — Bestlife73 · 2026-09-03