A consortium of German research institutions has released Soofi S, a 30-billion-parameter open-source language model that ranks first among fully open models on both English and German benchmarks. Trained on renewable-powered infrastructure in Munich, the model uses a hybrid architecture that maintains fast generation speeds even at very long contexts, and was deliberately weighted toward German-language data to serve European applications.
Summaries like this, in your inbox every morning.
Sign up free →What happened
A German research consortium released Soofi S, a 30-billion-parameter open language model trained entirely on Deutsche Telekom's Munich cloud facility. According to its pretraining report, Soofi S achieves the highest scores on both English and German benchmarks among fully open models, surpassing OLMo 3 32B and Apertus 70B.
Why it matters
The model uses a lean hybrid architecture that activates only about 3.2 billion parameters per token, giving it compute costs closer to a 3-billion-parameter model while maintaining throughput at very long contexts where conventional dense models drop off sharply. At 40,000 tokens with 32 parallel requests, Soofi S generates roughly eight times more tokens per second per GPU than dense models in the 14 to 24 billion parameter range. The deliberate emphasis on German-language training data—rising from 7.2 percent in the first training phase to 15.3 percent in the second—positions it as a strong alternative for European applications without reliance on non-European infrastructure.
What to watch
The consortium is seeking industry partners for the next phase to test the model in applications involving technical documents, code generation, and agent-based systems. Model weights, intermediate checkpoints, training code, and a detailed data inventory have been released; about 99 percent of the training mix can be independently reconstructed, though the exact license for the model's release has not yet been finalized.
Soofi S represents a deliberate effort by European institutions to build competitive open-source AI infrastructure independent of non-European providers. The consortium—coordinated by the German AI Association and funded by the German Federal Ministry for Economic Affairs and Energy as part of the European IPCEI-CIS program—includes the Fraunhofer Institutes IAIS and IIS, the German Research Center for Artificial Intelligence (DFKI), TU Darmstadt, the University of Würzburg, the L3S Research Center, the Berlin University of Applied Sciences, and companies Ellamind and Merantix Momentum. The model's training on sovereign infrastructure in Munich, powered entirely by renewable energy, signals a commitment to reducing dependence on externally controlled cloud providers.
The hybrid architecture choice has concrete practical implications. By combining Mamba-2 layers (which have superior memory efficiency) with selective use of standard attention, Soofi S avoids the linear memory growth that makes long-context inference expensive for conventional transformers. The benchmark results bear this out: while models like Apertus 70B and Qwen3 32B see sharp throughput drops as context length increases, Soofi S maintains nearly flat performance from 4,000 to 256,000 tokens. This efficiency appears valuable for industrial applications involving lengthy technical documents—exactly the use case the consortium now plans to validate with partners.
The training data composition reveals a strategic trade-off. By weighting German-language content far more heavily than Nvidia's reference recipe (15.3 percent versus 5 percent for all non-English languages combined), the consortium achieved top-tier German benchmark performance without sacrificing English capability. However, the model does show specific weaknesses: it scores significantly lower than Qwen3.5 on German math benchmarks (56 points on Minerva MATH-DE versus 76.5 for Qwen3.5) and exhibits a notable failure mode on the RULER extraction task at very long contexts (3 percent hit rate beyond 32,000 tokens versus 60–64 percent for Nemotron). The authors attribute the extraction weakness to a lack of synthetic long-context extraction data in the training set, identifying a concrete direction for future improvement. Regarding openness, Soofi S meets the Open Source Initiative's Open Source AI Definition 1.0, though it does not satisfy a stricter European standard that would require all training tokens to be freely distributable; the 1.3 percent share of commercially licensed Genios newspaper data prevents full alignment with that stricter threshold.
No comments yet. Be the first to share your thoughts!
Log in to join the discussion
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
