Search papers, labs, and topics across Lattice.
The paper introduces Soofi S 30B-A3B, a sovereign, open-source Mixture-of-Experts (MoE) foundation model designed for both German and English, which activates only 3B of its 30B parameters per token to enhance throughput for long-context applications. This model, pretrained on approximately 27 trillion tokens with a focus on German, achieves competitive performance against dense models while outperforming all European sovereign baselines, including larger models in active parameters. Soofi S sets a new standard for open models by achieving the highest evaluation scores in both languages among its peers, and it will be released under permissive open-access terms.
Soofi S outperforms larger European models while maintaining a fraction of the active parameters, redefining the potential of open-source foundation models.
We present Soofi S 30B-A3B, a sovereign, open-source Mixture-of-Experts (MoE) hybrid Mamba Transformer foundation model for German and English. Its hybrid design activates only 3B of 30B parameters per token and keeps the inference cache near-constant as context grows, giving it a decisive throughput advantage over dense models for long-context, high-concurrency deployment. Pretrained on roughly 27 trillion tokens with deliberately up-weighted German, Soofi S matches dense 14 to 27B models on aggregate English and German benchmarks while achieving the best code aggregates in both languages among 17 open base models, and outperforms every European sovereign baseline in our comparison, including ones far larger in active parameters. Among fully open models, Soofi S obtains the highest English and German evaluation scores, ahead of Olmo 3 32B and Apertus 70B. Soofi S was built end-to-end on the German Industrial AI Cloud, a sovereign HPC scale AI infrastructure operated by Deutsche Telekom in Munich. Soofi S will be released under highly permissive, open-access terms: weights, selected intermediate checkpoints, full per-source data accounting, hyperparameters, and training and evaluation code. Where source licenses permit, data-construction artifacts are released under permissive licenses; commercially licensed sources are documented with aggregate statistics and exact mixture accounting.
