Live Feed
Qwen3.8-Omni-Flash: 1M Context, Four Modalities, Audio Surpasses Gemini 3.8 Flash
Alibaba releases Qwen3.8-Omni-Flash, a native omnimodal model accepting text, image, audio, and video in a 1M-token context. Overall audio performance exceeds Gemini 3.8 Flash; API price per audio hour drops 98% from the prior generation.
Crusoe Raises $3.9B at $30.9B Valuation as Jane Street Deal and IPO Talks Emerge
Crusoe closed a $3.9B Series F co-led by Atreides, Mubadala, and Valor at a $30.9B valuation, with Nvidia, Qatar Investment Authority, and Founders Fund participating. A $13B five-year compute contract with Jane Street and IPO conversations with Goldman Sachs and Morgan Stanley underpin the raise.
OpenAI Configures GPT-6 Astra for Legal Research, Scoring 54% vs 38.7% for Base Model
Astra for Law combines GPT-6 Astra with a 230-million-URL legal index and 99.9% of U.S. precedential case law. On the Vals AI Legal Research Bench, it passes 54% of questions versus 38.7% for the base model with web search alone.
Plugin4Shell: Zero-Click RCE Found in Claude Code, Codex, GitHub Copilot, and Gemini CLI
Air Security discloses a SHA-pinning bypass that enables zero-click remote code execution across every major AI coding agent. One design flaw, independently repeated by four labs, leaves millions of developer machines exposed.
DeepMind Institute Maps 11 AGI Economic Interventions Using 51 AI Agent Economists
A new essay from Google DeepMind's research institute evaluates 11 household-facing policy responses to AGI-driven economic disruption using a novel methodology: 51 AI agent raters modeled on real economists. The framework links different interventions to different AGI scenarios rather than backing any single universal policy.
AI Wins the Metaculus Cup for the First Time, Beating Rivals That Spent $15M Combined
On September 5th an AI took first place in the Metaculus Cup forecasting competition, with other AI systems placing second and fifth. Separately, Cassi AI became the first system to outrank human superforecasters on market questions in ForecastBench.
Fujitsu's MONAKA CPU: 2nm, 3.8GHz, Claims 2x AI Inference Throughput with Half the Server Count
Fujitsu begins global sales of the MONAKA CPU in November 2026. Built on a 2nm 3D-stacked process and designed for sovereign AI deployment, the chip is pitched as halving AI inference server requirements versus competing CPUs.
Hyperscale Data Has Spent $70M to Ready Michigan's 340 MW AI Campus — Operations Start November
Hyperscale Data says it has acquired the equipment needed to bring its Michigan AI data center online in November 2026 under a 20-year master services agreement. The campus holds 340 MW of total potential capacity, with the initial 20 MW now six weeks from first revenue.
Z.ai Says GLM Built Its Own Inference Infrastructure on 100,000 Chinese Chips — An Early RSI Signal
Z.ai deployed GLM-5.3-Flash on production infrastructure its own model helped engineer, running on more than 100,000 Chinese-made AI accelerators. The lab calls it an early form of recursive self-improvement and says the model is moving toward replacing its own engineers.
Artificial Analysis Triples Terminal-Bench Weight in Engineering Index, Upgrades Six Domain Benchmarks
Artificial Analysis has released Capability Indices v1.1, raising Terminal-Bench's weight in the Engineering index from 5% to 15% and adding AutomationBench-AA to Finance, Legal, and Strategy scores. The update rebalances domain AI rankings away from pure reasoning toward agentic execution.
Cloudflare Open-Sources the 6-Phase Security Audit Agent Its Engineers Use
Cloudflare has released the AI security audit tool its own team uses internally. The agent runs six sequential phases across a codebase, mapping trust zones and confirming exploitability before surfacing results, and slots into existing coding agents like Claude Code, Cursor, or Codex.
Yotta Commits $12B to 80,000-90,000 Vera Rubin and GB300 GPUs in One of the World's First HBM4 Orders
Indian data center operator Yotta is betting 12 billion dollars on up to 90,000 Nvidia GPUs spanning the Vera Rubin and GB300 generations. The order locks in India's largest GPU operator ahead of a planned IPO targeting a 6 billion dollar valuation.
A 4B Model Trained with RL Produces Query Plans 81% Faster Than Postgres on Join-Heavy Workloads
Rohan Bansal fine-tuned a 4-billion-parameter model using supervised learning and agentic reinforcement learning to generate Postgres query plans that run 1.81x faster per query (geometric mean) than the database's own optimizer. The model started unable to produce a valid plan for 99 of 113 test queries.
Arena's HarnessTax Study: Coding Agent Harness Choice Moves Cost, Not Success Rate
An Arena.ai paper evaluating 7 models across Claude Code, Codex, and Pi finds that the choice of coding harness has surprisingly little impact on task success but can swing costs significantly. Claude Fable 5 reaches 97.8% on SWE-bench Lite regardless of which harness wraps it.
Nvidia Launches Two Official Rust Tracks for Writing CUDA GPU Kernels
Nvidia has published cuda-oxide and cutile-rs, giving Rust developers two distinct paths to write GPU kernels: one based on the familiar SIMT thread model, one on the newer Tile abstraction. Neither requires dropping into C++.