2+ day, 13+ hour ago   (115+ words) According to KyeGomezB, DeepSeek’s causal encoder‑decoder MoE cuts active params to 8B in, 16B out, promising lower cost and top benchmarks. By activating fewer parameters during input stages, the model reduces memory usage and speeds up initial processing. Output stages then…...

4+ day, 13+ hour ago   (208+ words) According to DeepLearning.AI, Zai’s GLM-5.3-Flash hits 320B params, runs China-made hardware, and leads GDPval AA v2 at $0.09 per task. The GLM-5.3-Flash model from Zai represents an important step forward in efficient large language model design. With 320 billion parameters yet…...

1+ week, 3+ day ago   (110+ words) Tencent: Hy4 Preview Lands on B.AI API blockchain.news Tencent: Hy4 Preview Lands on B.AI API Tencent Hunyuan 770B Hy4 Preview model with 49B active parameters opens on B.AI API after 203 internal tests. Justin Sun flagged the Hy4 Preview launch on B.AI API, delivering Tencent Hunyuan's…...

2+ week, 1+ day ago   (76+ words) GLM-5.3: Z.ai Releases Open-Weight Coding Model blockchain.news GLM-5.3: Z.ai Releases Open-Weight Coding Model GLM-5.3 tokens surpass one billion in personal use, with AveryChing highlighting strong coding throughput and value on the new open-weight release from Z.ai. AveryChing burned through…...

2+ week, 2+ day ago   (442+ words) Nvidia Expands Chinese open models amid crackdown warning blockchain.news Nvidia Expands Chinese open models amid crackdown warning According to @CNBC, Nvidia boosts support for Chinese open AI models while warning of tighter White House export crackdowns impacting data center…...

2+ week, 4+ day ago   (209+ words) Peter Zhang Aug 25, 2026 18:18 SkyRL's FP8 reinforcement learning stack matches BF16 convergence while reducing rollout step time by up to 23%, boosting efficiency on NVIDIA GPUs. These optimizations are particularly impactful for LLM reinforcement learning, where memory bottlenecks often limit performance. For example, SkyRL’s…...

2+ week, 5+ day ago   (379+ words) Alibaba Accelerates AI with $10.2B war chest blockchain.news Alibaba Accelerates AI with $10.2B war chest According to @CNBC, Alibaba raised $10.2B to fund full stack AI, pledging all proceeds to infrastructure and model capabilities, per CNBC. Alibaba announced an 80 billion Hong Kong…...

2+ week, 5+ day ago   (279+ words) According to @CNBC, Alibaba will place $10.2B in shares to fund cloud AI, model training, and chips, as shares drop on dilution concerns. Alibaba announced a major $10.2 billion share placement on August 24 2026 to accelerate its artificial intelligence initiatives according to CNBC....

2+ week, 5+ day ago   (116+ words) Alibaba: Plunges on $10.2B Share Placement for AI blockchain.news Alibaba: Plunges on $10.2B Share Placement for AI Alibaba shares plunged after $10.2 billion placement to fund AI push, with immediate Hong Kong market reaction and dilution pressure on the tech giant. Alibaba…...

2+ week, 6+ day ago   (357+ words) According to FoxNewsAI, Fetterman supports faster US AI development and warns China could gain if regulations overreach, as reported by Fox News. Senator John Fetterman aligned with President Trump on accelerating artificial intelligence development in the United States on August…...