China AI

AITencent has released a preview of its Hy Image 3.5 image generation model, which product teams across Yuanbao, WorkRally, Ima, and other services are helping refine through co-design.

AITencent Hunyuan has made its Hy Image3.5 preview available on the OnSolo platform, free for two weeks.

AIShanghai AI Lab and SJTU's LUMIA Lab released NCP-ArchPreview, an 8.9B open-weight language model under Apache 2.0.

Why it matters: The post pairs an Apache 2.0 open-weight release with training-efficiency figures, showing how the concept module adapts to new domains with few trainable parameters.

AIWorkBuddy is offering free access to DeepSeek-V4.1-Flash until October 9th. The post invites users to apply the model to their tasks while the offer lasts.

AITencent Hunyuan's Hy Image3.5 preview is now available in ComfyUI, with a claimed 30% higher win rate in human evaluation than Hy Image3.0.

AIinclusionAI open-sources the Ming-Image-0.1-Design family, two complementary 6B models for visual-design workflows, under an MIT License.

Why it matters: The post separates a design-generation model from a layer-decomposition model, letting users compare two distinct visual-design workflows under one MIT license.

AIAlibaba's Qwen-Image-2.1 is now the top open-source model in both the Image Edit Arena and Text-to-Image Arena, according to Arena.

AIAnt Ling released Ling-3.0-flash-fin, a finance-specialized open-weight model with 124B total parameters and 5.1B activated, offering high intelligence density for its size.

AITinker says an open LLM can serve a Jev-like interface that takes discrete options and returns fast probabilities, since next-token prediction is already a probabilistic classifier.

AIUnsloth Desktop received a hotfix update adding image editing for Qwen-Image-2.1.

AIUnsloth announced that Qwen-Image-2.1 FP8 and GGUF quantized versions should now run properly in Unsloth Desktop.

AIFireworks Research released Ember-1, a specialized model built on Kimi K3 that it says delivers the same quality with 40% fewer tokens.

Why it matters: The source gives benchmark and A/B results for cutting reasoning tokens while holding quality, which bears on cost planning for coding and agent workloads.

AIComfyUI's update makes the MiniMax H3 video VAE encode up to about 2.2x faster and decode 1.4-2.7x faster, cutting a 1344x768, 129-frame round trip on an RTX 5090 from 24.3 to 12.7 seconds.

AIStepFun has open-sourced onPanda, a tool used internally for LLM data annotation and model inspection, letting users correct tokens and let models continue.

AIStepFun's Step Code passed 72 of 89 tasks (80.9%) on Terminal-Bench 2.1, tying for the highest pass rate among evaluated harnesses while using fewer tokens than the other tied leaders.

AIStepFun has released Step Code v0.1.0, an open-source command-line tool under the MIT License that covers reading and editing code, running tests, and shipping from one CLI. The post reports 80.9% on Terminal-Bench 2.1 and 73.3% on Multi-Frame, a 150-task long-horizon benchmark from StepFun.

AIDaniel Han says Qwen-Image-2.1 works in Unsloth Desktop through INT8, FP8, and GGUF builds, with Unsloth also releasing dynamic GGUFs for it.

AIUnsloth says the 7B Qwen-Image-2.1 text-to-image and editing model can run locally on 12GB VRAM using its GGUF builds.

Why it matters: The post gives concrete local-run settings, VRAM figures, and GGUF and FP8 options, which helps readers judge whether the model fits their hardware.

AIChinese authorities are investigating allegations from Anthropic that AI firms including Moonshot and DeepSeek may have facilitated leaks of sensitive Chinese military, police and state-owned corporate data to the U.S. The image text says the Cyberspace Administration of China summoned representatives of the seven companies named in Anthropic's report and later focused on DeepSeek and Moonshot, with officials interviewing executives and employees at their offices.

AIXiaomi's MiMo-V2.6-Pro ranks first among open-weight models on the Artificial Analysis Intelligence Index with a score of 46.

AIJS Denain of Epoch AI discusses recursive self-improvement, arguing public evidence does not yet show a software intelligence explosion, though OpenAI's reported 2X monthly growth in researchers' Codex spending suggests substantial value.

AITencent Hunyuan has released a preview of Hy Image3.5 on Miora, a design platform, and is offering it free for two weeks.

AIT-Head, Alibaba's chip subsidiary, unveiled the Zhenwu V900 AI chip for training and inference at the 2026 Apsara Conference in Hangzhou.

AIKimi has released its Kimi Browser Extension, formerly Kimi WebBridge, which runs in the browser sidebar to navigate websites and fill out forms.

AITencent Hunyuan posted a set of showcase images generated by its Hy Image3.5 preview model. The post presents the outputs as examples of the model's capabilities, without accompanying detail on methods or benchmarks.

AIAlibaba is preparing its Zhenwu V900 proprietary accelerator for commercial release in the first quarter of 2027, claiming three times the performance of the M890.

AIOpenBMB thanked a developer for testing MiniCPM5-2B as a worker in a multi-agent workflow handling invoice matching, short payments, duplicate references, and disputes through tool calls.

AIMoonshot's Kimi K3 is now available on Amazon Bedrock, with its license requiring a paid agreement for model-hosting businesses and affiliates above $20M in annual revenue.

AIAlibaba CEO Eddie Wu said on September 22 that strong AI demand is driving infrastructure investment, targeting over 20GW of global cloud data-center capacity by 2032.

AIQwen-Image-2.1 is now ready to run optimized on Intel hardware with Day-0 OpenVINO support, announced by Intel's developer account. The open-weight checkpoint handles both image generation and editing within a single model.

AIMoonshot AI's Kimi K3 is now available on Amazon Bedrock for coding, document analysis, and extended agent workflows. Bedrock provides access, encryption, and auditing controls, and explicit prompt caching is supported.

AIStepFun's Step 5 Preview scores 44 on the Artificial Analysis Intelligence Index at about $0.72 per task, matching Kimi K3 (max) at roughly 2.8x lower cost.

AITencent Hunyuan has launched a preview of Hy Image3.5, which it says wins 30% more often than Hy Image3.0 in human evaluation.

AITencent Hunyuan says it shrank the 770B-parameter Hy4 preview from roughly 1.5TB to 214 GiB while keeping the parameter count unchanged.

Why it matters: The quoted Zhihu post explains how Hy4 preview's weights were quantized and kept usable at inference, a concrete engineering case for compressing large MoE models.

AIA post praises Xiaomi for matching frontier-level results after only 30 reinforcement learning steps. A quoted reply notes the model improved dramatically in those 30 steps, with no sign of a plateau, and wonders why training stopped there.

AIArena says Xiaomi's MiMo-V2.6-Pro debuted at about #10 overall on Code Arena: WebDev with a 1628-point AutoEval score, tying Claude Fable 5 (High).

AIXiaomi's MiMo-V2.6-Pro UltraSpeed is now available in MiMo Desktop and through the API, delivering up to 20× faster generation.

AIXiaomi MiMo announced open-source releases of Pro and Flash, the MiMo-V2.6-Distill-Qwen-9B model, a technical report, over 7K RL task environments, an end-to-end RL framework, and composable mini-harnesses.

Why it matters: The table compares a 9B distilled model against Qwen3.5-9B on coding, cyber, and agent benchmarks, showing how the reinforcement learning stage changes results.

AIXiaomi's MiMo-V2.6 combines code, design, and tool use to build frontend interfaces, presentations, Figma-linked visual assets, and videos.

AIXiaomi says MiMo-V2.6 keeps API pricing unchanged from V2.5 for both the Pro and Flash models while adding more intelligence.