AI2 open-sources Olmo-core 3, a MoE training stack scaling to trillion parameters
lmoroney · x · 2026-10-03
The Allen Institute for AI open-sourced Olmo-core 3, a PyTorch training stack for MoE models up to the trillion-parameter range.
- In one test, growing the expert pool from 8 to 128 while keeping active parameters around 3.2B dropped training throughput by less than 5%.
- On eight B300 GPUs it reports 2.7x the tokens-per-second of their earlier FSDP-based MoE path.
- Apache-2.0 on GitHub (allenai/OLMo-core) and on PyPI as ai2-olmo-core.
- A useful open case study in expert parallelism, routing costs, and keeping sparse models efficient as they grow.
Related event: Ai2 Open-Sources Olmo-core 3, a Trillion-Parameter MoE Training Stack(5 posts)→
More from Infra
- Gemma.c: full Gemma 1/2/3 multimodal inference in a single C file, ~72 tok/s on CPU at INT8 — This-Purchase-3325 · 2026-11-16
- Sarvam AI Launches Token Factory: India-Hosted Inference for Open Models — itsOmSarraf_ · 2026-11-16
- BIS Issues Interim Final Rule Addressing Remote Access to Advanced AI Chips — apples_jimmy · 2026-11-16
- Jailbroken PS4 runs a 9B open-source model at up to 43 tok/s by talking directly to its 2013-era GPU — AIFlow_ML · 2026-11-16
- Mobile AI Factory: agents auto-build on-device SLMs with 19x less GPU time — pmttyji · 2026-11-16
- vLLM serves DeepSeek V4.1-Flash at 954K tokens/s per GPU, $38.96/GPU-hour on GB300 NVL72 — bookwormengr · 2026-11-16