Leaks and official confirmations keep Qwen at the center of open local AI, including 120B-class models, embedded NPUs and long-context agents built on Qwen weights.
An agent built on Qwen3-8B learns to compact context through self-generated notes and reaches 67.8% accuracy on book question-answering benchmarks with books up to 2.5 million tokens, never holding more than 20K tokens. It beats a GPT-5.5 agent and n...
A pull request in the QwenLM/qwen-code repository shows unannounced model IDs: qwen4-max, qwen4-coder-plus, and qwen4-omni-120b-think, all resolving to a 262,144-token context window. The omni fixture detail confirms the models are not hypothetical, ...
Qwen's head Liu Dayi Heng has denied rumors that future Qwen series would move to API-only releases: Qwen 4.0 and 5.0 will remain open source. With 3 billion downloads, over 460 models, and 300,000 derivatives, the decision strengthens a strategic ec...
A researcher compiled an NPU into Qwen 3.5-4B weights and ran stories260k without external runtimes. QwenPU redefines the boundary between model and platform: weights become a computational substrate, with implications for self-hosted deployment, aud...
QwenPU compiles an NPU into Qwen 3.5-4B's weights to run Karpathy's stories260k, with no external tools and bit-exact operations. The project redraws the boundary between model and compute platform, with implications for self-hosted stacks and data s...
A detail in a pull request reignites the debate on a class of 120-billion-parameter LLMs. For teams managing self-hosted deployments, the size reshapes calculations around VRAM, quantization, and TCO, without official announcements. The signal points...
A pull request in the QwenLM/qwen-code repository has drawn attention to a possible 120-billion-parameter model. It is not an official announcement, but the 120B class touches a central issue for on-premise deployments: aggregated VRAM, quantization,...
