today 11:05Isomorphic Labs Raises $2.1B Series B for AI Drug Discovery; Thrive Capital Leads with UK Sovereign AI Fund Among New Backers
today 11:05White Circle Raises $11M Seed for Real-Time AI Model Policy Enforcement Layer; Backers Include OpenAI, Anthropic, Mistral, and Hugging Face Founders
today 11:05Legora Extends Series D to $600M; Nvidia and Atlassian Join at $5.6B Valuation as Legal AI Startup Crosses $100M ARR
today 11:05AgentEscapeBench (arXiv:2605.07926): 270-Instance Escape-Room Benchmark Finds Best LLM Agent Drops from 90% to 60% as Tool Dependency Depth Increases
today 11:05arXiv:2605.06241: RL for LLM Reasoning Touches Only 1–3% of Token Positions; RL-Free ReasonMaxxer Matches Full RL Training at 1,000× Lower Cost
today 11:05arXiv:2605.06638: ScaleLogic Framework Shows RL Training Compute Follows Power Law With Reasoning Depth; Scaling Exponent Rises from 1.04 to 2.60 as Logical Expressiveness Increases
today 11:05CreativityBench (arXiv:2605.02910): 14K Affordance-Based Creative Reasoning Tasks Show LLM Scaling Quickly Saturates; Chain-of-Thought Provides Minimal Gains
today 11:05GazeVLM (arXiv:2605.07817): 4B VLM Adds Trainable Gaze Tokens for Dynamic Causal Attention Control; Beats Comparable VLMs by ~4% and Agentic Pipelines by >5% on HRBench
today 11:05Delivery Hero Deploys Herogen: In-House Coding Agent Reaches 130-Engineer-Equivalent Output, 85% PR Merge Rate, 100+ PRs per Day
today 11:05Z.ai Releases GLM-5.1 Open-Source: 754B MoE Tops SWE-Bench Pro at 58.4, Surpassing GPT-5.4 and Claude Opus 4.6; MIT License