PayPal is hosting a global online hackathon with $69,750 in total prizes, inviting developers and builders to create projects that meaningfully integrate PayPal's payment platform with AI tools. Participants can build agents, apps, automations, or new payment experiences, submitting working prototypes with code, demos, and documentation for evaluation by PayPal judges.
Rust 1.99.0 introduces language stabilizations including C-variadic functions and naked function attributes, expands linting capabilities for safety and correctness, optimizes RangeInclusive iteration, stabilizes new APIs for Box and Vec operations, and adds a new debug profile in Cargo for improved development workflows.
Rust 1.99.0 has been released, introducing support for C-ABI variadic functions defined in Rust, layout information retrieval from raw pointers, and numerous stabilized APIs. The release also updates documentation guidance on Box::leak to recommend against round-trip unleaking patterns.
Cloudflare launched a Monetization Gateway beta that enables domain owners to charge AI agents for access to websites, APIs, and data using per-request pricing and blockchain-based payments. The gateway addresses the mismatch between traditional subscription models and agent-driven consumption patterns by offering fast, low-friction payment rails via HTTP 402 status codes and stablecoin transactions.
Claude Code and similar AI tools operate by tokenizing text and generating responses one token at a time based on probability distributions, with no planning, memory, or knowledge beyond their training cutoff. The article explains how this architecture leads to predictable behaviors and common failure modes like hallucinations when models make incorrect assumptions about unseen code or use outdated APIs.
Cloudflare launched Application Profiles, a positive security tool that learns expected HTTP request structures from traffic analysis and blocks deviations, helping defend against AI-powered attacks by enforcing request conformity rather than just signature-based detection.
Holo4 is a new series of agentic models (27B dense and 35B-A3B MoE) that can interact with software through GUIs, code, APIs, and MCP tools across desktop, web, and mobile platforms. Trained via supervised learning and reinforcement learning on 10,000 tasks from an internal Agentic Task Factory, Holo4 27B scores 61.7% on OSWorld 2.0 at $0.08 per task, competing with frontier models at significantly lower cost.
Holo4 is a new series of agentic models (27B dense and 35B-A3B MoE) designed for computer-use tasks that interact with software through GUIs, code, APIs, and MCP. Trained via supervised and reinforcement learning on diverse environments, it scores competitively with frontier models on academic benchmarks like OSWorld 2.0 while operating at significantly lower cost and parameter count.