7B (hybrid RNN/Transformer dense)
Apache 2.0
Hybrid OLMo-family release with staged checkpoints covering Base, Think-SFT, Instruct-SFT, Instruct-DPO, and Think tracks; architecture alternates 3 gated DeltaNet layers for every 1 attention layer (75% DeltaNet, 25% attention), with 32 layers and 65,536 context length; model-card claims include roughly 2x data efficiency over OLMo 3 and major long-context throughput/memory efficiency gains; release includes training data details and published datasets for pretraining, mid-training, and long-context stages (Dolma3 and Dolmino/Longmino mixes).