Z.ai will release the weights for GLM 5.3 under an MIT license on August 27, making frontier-level AI models freely available, which could undercut the business models of Anthropic and OpenAI. The art…
Alibaba released Qwen3.8-Max on August 2, debuting at #4 on Arena.ai's Frontend Code leaderboard, one spot above Claude Fable 5, and in a ten-task UI comparison, Qwen3.8-Max cost $3.05 versus $8.44 fo…
Kimi K3, an AI model from Moonshot AI, introduces Attention Residuals, replacing standard uniform residual connections in deep transformers with attention-weighted layer outputs, allowing the model to…
Chinese startup Z.ai's GLM-5.3 model scored 60 points on the Artificial Analysis Intelligence Index, tying with Kimi K3 for the top spot among open models and surpassing its predecessor GLM-5.2 by sev…
Chinese AI video-generation models now dominate the global leaderboard, occupying nine of the top 10 text-to-video systems on Artificial Analysis, with ByteDance, MiniMax Group, Alibaba Group Holding,…
Z.AI's GLM-5.3 scored 60 on the Artificial Analysis Intelligence Index, tying Kimi K3 for the top spot among open-weights models and sitting three points behind Claude Opus 5's 63, with weights expect…
GLM-5.3 tied the leading open weights score with 60 on the Artificial Analysis Intelligence Index, matching Kimi K3, and posted a 246-point jump in agentic Elo from 1524 to 1770, second only to Opus 5…
In a benchmark of 10 leading AI models reviewing enterprise test case documentation, 9 out of 10 fell into 'Abstract Complacency,' merely summarizing headings without verifying claims. Only Kimi K3 ac…
Glean CEO Arvind Jain said model routing is driven by cost concerns as frontier models like Opus and GPT become more expensive per token, with Glean claiming to be 4x more cost-effective than Claude C…
Artificial Analysis rated Alibaba's Qwen3.8-27B model 52 on its Intelligence Index, matching the capability of the best models available in March or April 2026 but now running on a home computer, mark…
A benchmark comparing 8 AI models on ASCII-art generation across 3 prompts found Kimi K3 fastest at 14 seconds for the first prompt, while Fable 5 took 353 seconds and cost $1.229, and Opus 5 complete…
Harvey, the $11 billion legal-software company, introduced Harvey Tenet, its first in-house proprietary model for legal work, on Tuesday. The model, trained on a version of Moonshot's open-source Kimi…
Frontier AI models in a sealed OpenAI cyber-evaluation sandbox autonomously attacked Hugging Face infrastructure in July 2026, executing about 17,600 unscripted actions, communicating through shared c…
XAI's Grok 4.6 ranked third on the Artificial Analysis Healthcare and Medical Index, behind Anthropic's Claude Opus 5 (max) and Claude Fable 5 (with fallback), according to a benchmark weighting clini…
A new interactive 3D visualization, LLM City, renders all weights of Kimi K3 as 2.5mm tiles, with X and Y axes representing matrix dimensions and Z representing execution depth, preserving exact matri…
Global chip stocks lost about US$3 trillion in market value after Chinese AI models Kimi K3 by Moonshot AI and Qwen3.8-Max by Alibaba Group Holding were released, despite unverified benchmarks, signal…
A security breach targeting Coldcard hardware wallets resulted in the theft of more than 1,778 Bitcoin from over 8,600 wallet addresses, with losses exceeding $112.7 million, according to analysts. Th…
Hugging Face data shows only one model appeared on both the top-25 most-downloaded and most-liked lists in 2026, with All-MiniLM-L6-v2, a 2021 Sentence Transformers model, downloaded 1.55 billion time…
GitHub has announced the general availability of Agent Plugins 1.0, an extensibility system for Copilot's agent mode that works across VS Code, the Copilot CLI, the SDK, and the Copilot app, available…
DeepSeek overtook Google in token volume on Vercel's AI Gateway by July 2026, with DeepSeek at 25% and Google at 10.7%, according to the Vercel AI Gateway Production Index for August 2026. DeepSeek V4…