China AI

AIMagic says its new pretraining recipe matches DeepSeek V4 Pro's pretraining while using 50x less compute, roughly half the FLOPs used for GPT-3, or about $0.5M on GB200.

AIUnsloth's Qwen3.8-27B GGUF is now the most-liked GGUF ever on Hugging Face, with the post predicting it will soon enter the top 30 most-liked models overall.

AIOpen model licenses are tightening at the Chinese frontier, with Zhipu's GLM-5.3 switching from MIT to a custom license requiring a security review for inference and fine-tuning providers with over $10 billion in annual revenue.

AIUnsloth's Qwen3.8-27B GGUF has become the most-liked GGUF model of all time on Hugging Face, reaching 10 million downloads and 3.7K likes in 24 days. The post links the GGUF model page and a Qwen3.8 guide on the Unsloth docs site.

AIXiaomi MiMo has launched MiMo Desktop in invite-only beta, a desktop agent that turns Office files, images, video, audio, and zips into finished, editable output.

AIBaidu has launched AI, Evolving, a new podcast series, with its first episode examining AI's growing role in scientific discovery through Famou's work on pine wilt disease.

AITencent Hunyuan says its Hy4 preview has been upgraded to reduce long thinking and over-verification on complex tasks, which users had flagged.

AIOpenBMB released JustRL-II-base-model, the pre-RL starting checkpoint for the JustRL II math-reasoning case study, scoring about 61% on AIME 2025 before reinforcement learning.

AIOpenBMB released UltraData-Code-L2-Classifier, a suite of language-specific file-level scorers for 11 programming languages in UltraData-Code-L1.

AIOpenBMB released MiniCPM5-2B-DSpark, a 323,776,001-parameter DSpark draft checkpoint with five layers that proposes seven draft tokens per forward pass for the MiniCPM5-2B target model.

AIOpenBMB has released MiniCPM5-2B, a dense 2B Transformer built for on-device and resource-constrained deployment, with an average score of 53.9 in its comparison set.

Why it matters: The release pairs a 2B model with open training datasets and reports per-benchmark comparisons against named same-size and larger models, letting readers check the claims directly.

AIBaidu hosted Hong Kong's Secretary for Transport and Logistics, Mable Chan, and LegCo Panel on Transport members at Apollo Park in Beijing.

AIA Berkeley Sky Lab researcher says stronger open models and new inference systems will make powerful local AI practical.

AIUnsloth Desktop now runs GLM-5.3-Flash GGUFs out of the box with faster inference, enabling MTP and faster long-context decoding.

AIUnsloth reports that its optimized GGUF build runs GLM-5.3-Flash locally 1.6 to 3.4 times faster, using improved decoding and multi-token prediction.

AIBAAI and Peking University researchers released official weights for ConsiSpace, a geometry-consistent multimodal framework for spatial reasoning in long-form visual observations.

AITencent has open-sourced EVIE-8B, an 8.4B-parameter visual document retriever that scores 66.75 nDCG@10 on ViDoRe V3 and ranks first on that leaderboard's mean task score of 66.24.

AITencent released EVIE-4.5B, a 4.5B-parameter visual document retrieval model, with weights, training pipelines, HAC token compression, and evaluation suites open-sourced on Hugging Face.

AICongratulations to @HUMAIN and @MiniMax_AI on HUMAIN-M3, a frontier Arabic model now available on HUMAIN Node 🎉. @vllm_project is running the inference under the hood, and we're looking forward to more Arabic use cases with this state-of-the-art stack.

AIZhipu (Z.ai) reported first-half 2026 revenue of US$142 million, up nearly 400% year on year, driven mainly by its open platform and API business.

AIMirai is releasing speculative decoding in its Uzu inference engine, starting with Qwen3.6-27B. The quoted post reports 105 output tokens per second on an Apple M5 Max with 128 GB of unified memory, 2.9× faster than the fastest MLX speculative-decoding implementation Mirai benchmarked.

AIBaidu and IFAW have launched AI Guardian, a platform powered by ERNIE models that builds on a collaboration since 2020 that has helped remove more than 18,000 listings linked to the illegal wildlife trade.

AIBAAI released Recon2Reason-Reasoning-4B, a 4,437,815,808-parameter vision-language model fine-tuned from Qwen3-VL-4B-Instruct for indoor spatial reasoning.

AIBAAI and Hao Tang's team at Peking University's School of Computer Science released GaussianMind, an audio-language model built on RoboBrain and OpenMOSS Audio.

AIPrime Intellect reports that rebuilding RL weight transfer for GLM-5.2 on NIXL and ModelExpress cut sync time from 86.1 seconds with NCCL to 3.9 seconds in its fastest setting.

AIMeituan's LongCat-2.0 is now available free in Command Code, the AI coding tool. The model has 1.6T parameters, 48B activated, and a 1M-token context window. It is available on all Command Code plans to all subscribers.

AIUnsloth says MTP enables Qwen3.8-Flash-Next to run about 1.3 to 1.7 times faster at inference with no accuracy change.

Why it matters: The source gives concrete MTP speedup ranges, hardware memory requirements, and GGUF quantization sizes, which help readers judge whether local deployment fits their setup.

AIBaidu's latest AI Pulse highlights how its product portfolio, including DuMate, GenFlow, MeDo, and ERNIE Assistant, is built to take AI work from request to finished result.

AIMeituan's LongCat-2.0, a 1.6T open-weights MoE model with a 1M context window, is now free to use in Cline.

AIZhipu's Z.ai marks the first anniversary of the GLM Coding Plan by giving every current subscriber a Reset Card. The card refills both the weekly and 5-hour quotas, and usage can be checked on the personal or team plan pages.

AITencent Hunyuan says its Hy4 preview model found inference bottlenecks on its own and raised end-to-end throughput by 31.8% through operator fusion and communication optimizations.

AIOpenBMB released MiniCPM5-2B-Midtrain, a BF16 mid-training checkpoint taken before SFT in the MiniCPM5-2B series, on Hugging Face and ModelScope.

AIShanghai AI Lab's InternLM has published Intern Lumina U2, a 16B-parameter MoE model with 1B active parameters that handles text QA, image generation and editing, and image, video, and 3D understanding.

Why it matters: The model unifies text, image, video, and 3D understanding with image generation in one framework, a broader scope than single-modality releases.

AIDeepSeek introduces DeepSeek-V4-Flash-Vision-Exp, its first experimental multimodal model in the DeepSeek-V4 family, built on V4-Flash with visual modules.

Why it matters: The source compares the model with its text-only predecessor and Opus-4.8 on agent benchmarks, showing where vision gains occur and where text performance holds.

AIAlibaba NLP has released core-emb-8b, an MLLM-based multimodal embedding model that distills a reranker's compositional judgments to distinguish attribute-object bindings such as "a white plate and a black chair" versus "a black plate and a white chair." The 8B dense embedding model, built on the Qwen3-VL-based VL-Emb backbone, scores 0.666 total average on compositional benchmarks, 5.7 points above its backbone.

AIHelical Fusion CTO Wei Xishuo told the NFEC2026 AI-for-fusion forum that the company uses an autoencoder to compress hundreds of stellarator shape parameters into three.

AIAlibaba-NLP has published Core-Reranker-8B on Hugging Face, an 8B-parameter multimodal reranker fine-tuned from Qwen3-VL-Reranker to better distinguish attribute-object bindings in text and image relevance scoring.

AIAlibaba NLP has released core-emb-2b and core-emb-8b, multimodal embedding models built on Qwen3-VL that distill reranker judgments to better match attribute-object bindings in text and image retrieval.

AIAlibaba NLP released core-reranker-2b, a 2B-parameter multimodal relevance-scoring model built on Qwen3-VL-Reranker to better distinguish attribute-object bindings in text and image pairs.

AITencent Hunyuan has open-sourced Hy4 preview under Apache 2.0, a flagship mixture-of-experts model with 770B total parameters, 49B active per token, and a 1M context window.