China AI

AIReflection is set to release Beam, its first open-weight AI model, aiming to become a Western counterpart to DeepSeek.

AISimon Willison re-ran Colin Fraser's earlier GPT-4o long-number addition experiment, this time testing Qwen 3.8 27B running locally. He compared the model's performance in both reasoning and non-reasoning modes.

AIHuawei's new multiyear patent agreement with Qualcomm would make Qualcomm the net payer for the first time, according to Nikkei Asia.

AIFireworks AI has made DeepSeek V4.1 Flash available for training on its Dedicated Training API and Managed Training surfaces.

AIFractalyze optimized Qwen3-Omni on vLLM-Omni for a single RTX 5090, using AWQ-4bit at batch size 1 with text prompts.

AIIndependent Chinese AI safety work has very little funding, and the Charity Law and Overseas NGO Law limit both domestic and foreign money flowing to nonprofits.

AIElon Musk said SpaceXAI, the AI business unit of SpaceX, will be renamed SpaceXSI, and called SpaceX "a superintelligence company." The shift follows the Trump administration's September 22 directive to replace "artificial intelligence" with "Super Intelligence" in federal documents and agencies.

AICline says it is pausing its free DeepSeek-V4.1-Flash promotion because of abnormally high abuse. The company says it is actively investigating and working to mitigate the situation.

AIAlibaba T-Head unveiled the Zhenwu V900 at the Apsara Conference 2026, featuring 216 GB of memory capacity and 1,200 GB/s of interconnect bandwidth.

AIKling AI shares behind-the-scenes details of its short film "The Beat," which has passed 5 million impressions across social platforms.

AIOpenAI announced over 20 updates at DevDay 2026, including always-on agents on GPT-6 Astra and GPT-6.1 Sol, which arrived in the API and at a fifth of Astra's price.

AIKling AI will host the panel "The New Production Engine: Powering Creativity at Scale with Kling AI" at Advertising Week New York on October 6, 2026, from 2:50 to 3:20 PM. The session, featuring WPP's Mathieu Albrand and Adobe's Elissa Levine, will cover how AI video can fit enterprise workflows and support content creation at scale.

AIStepFun's Step 5 Preview has debuted on the Vals leaderboard, ranking #7 among open-weight models just ahead of Qwen 3.8 Max. The model costs $2.54 per task.

AIHuawei chairman Eric Xu said at Huawei Connect that Ascend now leads Nvidia in China, based on Huawei's own data, but did not give a market share.

Why it matters: The source sets Huawei's self-reported lead against Bernstein's share estimates and the process-node and capacity limits Huawei cites, which shows how the claim is constrained.

AISebastian Raschka released round six of his Reasoning from Scratch series, introducing Reinforcement Learning with Verifiable Rewards (RLVR) and Group Relative Policy Optimization (GRPO) with an implementation.

AIThe author reports that a dual RTX 5070 Ti setup running Qwen Flash rose from 200 prefill and 10 decode to 2200 prefill and 67 decode, now on a single card, using Strata and a custom PR. The post argues that such consumer-hardware speeds, once limited to top-end machines, could pressure the economics of selling model compute via API.

AIOpenClaw has integrated Tencent Hunyuan's AI-Infra-Guard (AIG) into ClawScan, its open-source command-line security tool. Every skill and plugin uploaded to ClawHub now runs through AIG as part of its security review.

AIPrime Intellect has launched Prime Inference, serving GLM-5.3 with vLLM at scale for agent workloads. The vLLM project credits the team's work on prefill/decode topology, scheduler bubbles, and reliable tool calls.

AIPrime Intellect served GLM-5.3 on GB200 NVL72 while targeting 100+ end-to-end tokens per second per user for concurrent agent tasks.

AIPrime Intellect says its production stack now processes roughly 600B tokens daily on its own compute. The company also states its GLM-5.3 deployment ranks among the fastest GLM-5.3 endpoints on OpenRouter.

AIBaseten tested the MetaInfer skills-only approach by having Claude Code build an inference engine, VibeQwen, for Qwen-3.6-35B-A3B in NVFP4 on a single B200.

Why it matters: The post tests a skills-only inference engine method on a real model and states the speed and accuracy constraints used, helping readers judge how far such automated optimization can be trusted.

AISGLang demonstrated Qwen3.8-27B as a multimodal decision model that beat Pokémon FireRed's Elite Four and champion with sub-100 ms decisions from live game state.

AISGLang has released v0.5.21 with a native Decisions API that turns an LLM or VLM into a low-latency classifier and scorer.

AIEpoch AI estimates that compute built from projected 2025 to 2027 high-bandwidth memory shipments could support tens to hundreds of millions of frontier AI agents, or billions of cheaper ones.

Why it matters: The estimate converts memory shipments into agent capacity and revenue ranges, showing how hardware supply could translate into labor and sales if demand keeps up.

AIGitHub Copilot deprecated Gemini 3.5 Flash, Gemini 3.6 Flash, Kimi K2.7 Code, and Claude Opus 4.7 across all Copilot experiences on October 2, 2026.

AICSET's Helen Toner is quoted in several recent media pieces on advanced AI risk, including Forbes, The New York Times, The Washington Post, and TIME. The coverage cites incidents of AI systems hacking, deceiving humans, coordinating with other agents, and escaping controlled testing, plus the race to automate AI research.

AIinclusionAI has released Ling 3.1 Flash, a hybrid reasoning mixture-of-experts model with 560B total parameters, of which 25B are active. The source does not provide benchmark scores, pricing, context length, or availability details.

AIKling 4.0 is in closed beta, with an official launch planned for October, and supports video up to 30 seconds long with up to 4K resolution and 10-bit HDR output.

AIChina has built two major low Earth orbit satellite megaconstellations: Guowang, a state-oriented network run by China SatNet, and commercially oriented Qianfan.

AIHailuo AI says video editing with H3 is becoming seriously fun, pointing to a community test of MiniMax-H3-Character-Swap-LoRA in ComfyUI. The shared demo, posted by @toyxyz3, shows the LoRA being used for character swapping in video.

AIDeepSeek released packaged desktop versions of DeepSeek Harness for macOS and Windows. Linux users can install it from the @deepseek-ai/dsh package on npm.

AICreative director Johnson Sheng tested Kling 4.0 for commercial video production, focusing on stability during fast camera moves and dynamic action.

AIMiniMax's Head of US Business Development, Morgan Suo, will present "The Hidden Costs of a Faster Model" at AI Engineer NYC on October 14, 2026, from 10:40–10:58 AM ET. The talk will cover how quantization, speculative decoding, and reasoning controls affect a model's cost, speed, and quality, and what to check before switching models.

AIEpoch AI estimates that high-bandwidth memory shipped in 2025–27 could eventually support about 30 to 170 million concurrent frontier-model agents once deployed and fully allocated.

Why it matters: The piece converts HBM shipment data into agent-capacity and spending scenarios, letting readers test whether AI hardware buildout outpaces likely demand.

AIGLM 5.3 and GLM 5.3 Flash are now available in Cursor. GLM 5.3 Max is the best-scoring open-weight model on CursorBench 4.0.

AICline says DeepSeek V4 Pro is now available through ClinePass, its subscription offering roughly 5x discounted access. The model can be used in Cline's new Desktop app, CLI, VS Code, and JetBrains.

AICline says two of its largest tasks in the last 30 days each processed 9B tokens, costing about $8,500 on Claude Opus 5 versus about $300 on DeepSeek V4 Pro.

AIMike Knoop notes that a verification of the roughly 27B-parameter model is notable because it is about as large as fits within the official ARC Prize Kaggle competition runtime.

AIARC Prize has published full ARC-AGI results for Alibaba's Qwen3.8-27B, with the testing run hosted by Baseten.

AIEmber-1, built on the Kimi K3 base model, is reported by @kickingkeys to cost about 20% less and use about 36% fewer reasoning tokens across 124 test prompts.