Radar / AI infrastructure / Ai2 Olmo-Core 3
Ai2 Olmo-Core 3: open MoE training stack
The Allen Institute for AI released Olmo-Core 3, a redesigned open training framework for mixture-of-experts models, benchmarked to 1.2T total parameters with under 5% throughput loss scaling experts 8 to 128, plus a technical report and Apache-2.0 code.
Why it matters
First fully-open training stack proven at trillion-parameter MoE scale — it tackles the expert-routing communication overhead that erodes MoE efficiency, and will underpin the next generation of open Olmo models.
What you could build with it
A university lab or indie research team could use Olmo-Core 3's expert-parallel primitives to train a domain-specific MoE (e.g., for legal or biomedical text) at a fraction of the usual engineering cost, since the routing, parallelism and FP8 kernels ship ready-made.
Does it hold up?
Too early to judge — released Oct 1; Ai2's own benchmarks are promising but no community training runs at scale have been reported yet.
Built with Ai2 Olmo-Core 3
- RuntimeWire: Ai2 opens MoE training stack after testing a 1.2T-parameter setupRuntimeWire · Independent analysis of the release, the DDP-vs-FSDP design change, and the caveat that the 1.2T figure is a systems benchmark, not a trained model.
Learn more
First spotted on article: source.
More AI infrastructure
Modular open-sources MAX inference server, Mojo stdlib and accelerator kernelsThe core of Modular's unified AI deployment stack is now open: the MAX inference server (OpenAI-compatible endpoints)…infra · JEV 0.76OmniRouteFree MIT AI gateway: one endpoint, 290+ providers and 500+ models with quota-aware auto-fallback.infra · JEV 0.75Microsoft releases 301,000 Copilot coding-agent tracesMicrosoft open-sourced 301,026 GitHub Copilot coding-agent sessions (9.3M model calls, 8.7M tool calls) with timings…infra · JEV 0.71Cloudflare Agents Week: Sandboxes GA, 50K concurrent Workflows, Managed OAuth for agentsA dozen agent-infrastructure launches in one week: persistent Linux Sandboxes (GA) with real shell/filesystem/state…infra · JEV 0.69ai-memory: long-term memory for agent coding CLIsRust solution for long-term memory for agent coding CLIs, facilitating handoff between different agents and sessions.infra · JEV 0.68DeepSeek open-sources Ascend infrastructure stackDeepSeek published Ascend-optimized versions of its NVIDIA-proven infra components — TileLang Ascend, DeepGEMM-Ascend…infra · JEV 0.65
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.