STORY · PRODUKTER_
Alibaba launches Qwen 3.5 series with up to 262K tokens context and on-device capabilities
Alibaba has released the Qwen 3.5 models in sizes from 0.8B to 9B parameters with built-in multimodality and long-context support up to 262K tokens. The models use a hybrid attention architecture (Gated DeltaNet) and are designed for edge deployment and lightweight agents, with demonstrations on iPhone 17 Pro and support in Ollama and LM Studio.
WHY IT MATTERS
The launch of efficient, compact models with extremely long context windows challenges the assumption that advanced AI requires massive models. This opens up local, on-device AI use without cloud dependency.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.