A developer has released Kev, a family of small decision-making models (0.8B, 4B, and 9B parameters) built on Qwen3.5, inspired by the architecture behind Jev. The models handle yes/no, multiple-choice, and rating questions within a single request and are compatible with TypeSafe's System One API, meaning developers can run Kev locally instead of relying on that hosted service. Full training code, evaluation data, and a web playground are included, and the models run on both CUDA and Apple Silicon hardware.
A screenshot circulated online showing the US Federal Register's search interface offering Alibaba Cloud's Qwen3:0.6B language model as a search option alongside the default semantic search. About a day after the screenshot spread, the Qwen options disappeared from the site, though an archived snapshot from September 16 confirms they existed.
US officials removed a Chinese-made Qwen AI search tool from the Federal Register website on Wednesday after social media users spotted it powering search of public regulatory comments. The tool, made by Alibaba, had been live for at least a day before removal, and neither the National Archives, the White House, nor the FBI has explained how or why it was deployed.
OpenJev is a browser-based experiment that runs a local AI model entirely on a user's own GPU, with no backend or waitlist required. It lets users compare how quickly a model can read probability scores for a set of allowed options versus generating the same distribution token by token, using models like MiniCPM5 2B or the lighter Qwen3 0.6B for smaller devices.
Following a rushed 'Lite' GGUF release four days after Qwen 3.8 27B launched, the team has now finished fully optimized ShapeLearn quantizations and benchmarked them against both the Lite versions and rival quants. The full models push the quality-versus-speed tradeoff further, with all five new variants topping the performance frontier across six GPU test configurations.
PrismML, a startup founded by Caltech researchers led by CEO Babak Hassibi, has released Bonsai 2 27B, a compressed version of Alibaba's open-source Qwen3.8 27B model that shrinks it from full size down to 5.9 GB. The compressed model retains about 98% of the original's benchmark performance, up from 95% in the prior Bonsai release, while being small enough to run on PCs and possibly high-end smartphones. The company has raised a $22.25 million seed round and counts Databricks co-founder Ion Stoica as an advisor.
A new ternary-quantized model, Ternary Bonsai 2 27B, has been released, built on Qwen3.8 27B and using {-1,0,+1} weights with FP16 group scaling to shrink the model to about 1.76 effective bits per weight and a 5.9GB footprint. Despite being over 9x smaller than its full-precision counterpart, it retains 98.2% of aggregate benchmark performance across reasoning, coding, vision and agentic tasks, and supports a 262K-token context window under an Apache 2.0 license.
Nari Labs announced its Qwen3-TTS and Qwen3-ASR models achieved leading positions on Coval's voice AI benchmark, which measures latency and word error rate for speech AI systems. The ASR Fast model hit 44ms latency with 3.6% error rate at $0.12/hour, while the TTS Fast model achieved 63ms latency with a best-in-class 3.8% error rate, undercutting competitors like AssemblyAI and Deepgram on price.
Tom's Hardware Premium's weekly roundup details extensive benchmarking of the Qwen 3.8 27B language model on consumer-accessible hardware, including an RTX 5090, Mac Mini, DGX Spark and Strix Halo systems, showing it can approach top-tier AI performance without cloud API costs. The issue also covers IFA show reports noting a market split between ultra-light MacBook-style laptops and pricey agentic AI PCs, squeezing out affordable mid-range machines, plus a note on Ajinomoto's role in supplying materials for the chip industry.
A developer outlines a workflow for running local large language models on an Apple MacBook Pro M5 with 48GB RAM, combining Ollama for model management, OpenCode as the development harness, and Docker's sbx sandbox tool for isolation. The setup uses models like Qwen 3.8 27B and Gemma 4 31B, pulled via Ollama and configured through a custom sbx kit for each project.
A developer profiling Apple's M3 Neural Engine found an RTL performance bug that cuts DRAM weight-streaming speed from a normal 45-60 GB/s down to just 17-19 GB/s whenever a model's total weight size lands on an exact multiple of 1 MiB, affecting 7 of the 15 models in the ANEMLL project. By rerouting around the flawed speculative prefetch path in the kernel's DMA engine, the developer restored full throughput, more than doubling token generation speeds for Llama 3.2 1B and Qwen3-8B.
A follow-up experiment tested four open models—DeepSeek V4 Flash, Inkling, Kimi K3, and Qwen3.8 A95B—by inserting the first 1% of GPT-5.5 Pro's reasoning trace into each model's own reasoning channel before letting it generate answers freely. Researchers then measured how much of GPT-5.5 Pro's visible answer text overlapped with each model's output. Qwen3.8 showed the largest jump, with overlap rising from 33.92% unprefilled to 54.50% with the GPT-5.5 Pro prefill, a 20.58 percentage-point increase, while other models showed much smaller shifts.
GoKawiil