Nvidia is expanding support for Chinese AI models including DeepSeek V4 Flash and Alibaba's Qwen 3.8, while warning that potential US restrictions on such support could materially affect its business.…
Alibaba's Qwen team released Qwen3.8-Flash-Next, an open-weight multimodal mixture-of-experts model previewing the architecture for the upcoming Qwen4. The model features 125 billion total parameters …
A developer built Vancine, a smaller OpenAI-compatible API focused on current Chinese frontier models from Qwen, Kimi, GLM, MiniMax, and DeepSeek, as an alternative to OpenRouter's broad catalog. The …
Qwen released Qwen3.8-Flash-Next, a 125B-parameter open-weight multimodal MoE model built on the Qwen4 architecture with a 262K context window, which outperforms Claude-4.6-Opus (Max) and can run loca…
A new arXiv study (2608.24901v1) finds that decodable 'empathy' directions in three instruction-tuned LLMs do not yield reliable control over automated empathy scores, with affective steering raising …
Alibaba's Qwen team released Qwen3.8-Flash, a multimodal model with a million-token context window, native support for OpenAI and Anthropic API protocols, and integration with developer tools like Cla…
The oh-my-agent project introduced a commit-msg hook that enforces exact Co-authored-by addresses for agent commits, ensuring correct GitHub attribution. The update also routes capabilities more stric…
Claude Opus 5 (Fast) topped the Newsroom Reliability v0.2 benchmark with a mean score of 0.73 at an estimated $0.0460 per task, according to a 50-task run by RuntimeWire. Z.ai's GLM 5.3 Flash, Google'…
Apple's M5 chip, announced October 2025, introduces a Neural Accelerator in every GPU core, making prompt processing 3.3 to 4.1 times faster than the M4, according to Apple's MLX benchmarks. Token gen…
A new control-arm trace from Niodoo's Hydrodynamic Swarm shows that Google's Gemma 4, with force_cap=0 and T=0, falls into a repetition well and emits a self-report of the failure—'Wait — It Is Happen…
Alibaba's Qwen research unit released Qwen3.8-Flash, a multimodal model that improves coding and office-task performance while reducing training costs, according to a statement published on social med…
Ollama has reintroduced an integration with Claude Desktop, enabling users to connect Anthropic's app to open-weight models such as Qwen, DeepSeek, and Kimi, after its first effort stalled. The integr…
Alibaba's Qwen organization activated the Qwen3.8-Flash API on QwenCloud on August 26, 2026, pricing it at $0.16 per million input tokens and $0.47 per million output tokens with a default 1-million-t…
Alibaba's Qwen team released Qwen3.8-Flash-Next, an open-weight multimodal Mixture-of-Experts model with 125B backbone parameters and 6B active per token, previewing the Qwen4 architecture. The model …
Perceptron AI, a Bellevue, Washington-based startup founded in late 2024, launched its Perceptron Mk1 vision-language model on May 12, claiming it processes factory video at 80-90% lower cost than com…
Alibaba's Qwen team released Qwen3.8-Flash-Next, a mixture-of-experts model that activates 6 of 125 billion parameters per token, claiming one-ninth the training cost of competitors while outperformin…
A user running Qwen 3.8 27B on a 3090 reports that switching to ninfer, a Qwen-focused inference engine, doubled their token generation speed from 35-40 to 55-60 tokens per second. The user is conside…
Arduino has announced VENTUNO Q, a standard Ubuntu computer with up to 40 dense TOPS of AI compute, supporting models like Qwen 3 4B, Qwen 2.5 7B, Gemma 4, Whisper, and YoloX. The board features a dua…
Forum users on Level1Techs discuss concerns about Chinese open-weight AI models like Qwen and Kimi, with some arguing that economic pressure from these models could bankrupt Anthropic and OpenAI, whil…
Apple Inc. announced the M6 Mac mini and M5 Ultra Mac Studio, aimed at local AI workloads and Mac clustering, with preorders open and most versions shipping September 22. The M6 is Apple's first 2nm c…