AI2 open-sources Olmo-core 3, a PyTorch MoE training stack with 2.7x throughput gains
lmoroney · x · 2026-10-03
Allen Institute for AI released Olmo-core 3 (Apache-2.0), a PyTorch training stack for MoE models up to trillion-parameter scale. Growing the expert pool from 8 to 128 at 3.2B active params cost under 5% throughput; on eight B300 GPUs it delivers 2.7x tokens/sec vs their earlier FSDP-based MoE path. A useful open case study in expert parallelism, routing costs, and sparse-model efficiency.
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