Ai2
6 published articles
Open Source
Ai2 cracked open every drawer in the AI cabinet, here's what's inside
Ai2 releases Olmo 3, a fully open model family from 7B to 32B, including training data, code, and tools. The release emphasizes transparency across the entire model lifecycle, from pretraining data to post-training pipelines, setting a new standard for open AI research.
2026-07-11
AI Open Source
Two AI labs just proved why open models win in regulated industries
Domyn and AISquared leveraged Ai2's open-source Olmo family and Dolma/Dolci datasets to create models for regulated industries. Full transparency into architecture and training data helped them meet compliance requirements and cut infrastructure costs by roughly 50%.
2026-07-11
Artificial Intelligence
Ai2's EMO: the MoE model where modularity emerges from data, not human priors
Ai2's new MoE model, EMO, uses a novel training method that lets expert modules emerge naturally from data, enabling selective expert use with minimal performance loss. The model matches standard MoE performance on benchmarks while offering vastly improved modularity.
2026-07-07
Open AI Infrastructure
Ai2 just opened an AI cluster that publicly shares every mistake, not just the win
Ai2 just turned on a new AI cluster under the NSF OMAI project, one that makes models, tools, and processes fully open. An internal study found 82% of training effort goes to dead ends. Every GPU hour counts more when it fuels research across language, multimodal, and scientific domains.
2026-07-06
Model Evaluation
Ai2's olmo-eval gives LLM developers a microscope for every checkpoint
Ai2's olmo-eval brings per-question diffs and modular benchmarks to active LLM development, helping researchers tell real progress from statistical noise.
2026-07-06
Artificial Intelligence
Ai2 cuts satellite imagery AI costs by 3x with a smarter token trick
Ai2's OlmoEarth v1.1 reduces compute costs by up to 3x compared to v1, enabling cheaper large-scale map refreshes. The key innovation is merging resolution-based tokens for Sentinel-2 imagery, cutting token counts by a factor of three while preserving performance through modified pre-training.
2026-07-05