Agents & InferenceHacker News
German AI consortium releases Soofi S, an open 30B model that tops benchmarks
Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.
Summary A
Soofi S delivers 30B-level accuracy with only 3.2B active parameters per token, slashing inference cost and GPU memory by ~10× while keeping latency flat even for long contexts. This lets you serve high-quality German/English models on 24 GB GPUs or cut cloud spend by an order of magnitude without sacrificing throughput or accuracy.
Summary B
A 30B language model, Soofi S, activates only 3.2B parameters per token, achieving top benchmark scores while keeping compute costs comparable to a 3B model; this enables shipping large-capability models at lower inference costs, changing the cost-capability tradeoff for production LLM deployments.
0 picks