World Wire

The vLLM Semantic Router team has released Decision 3.0 , a family of multimodal decision models. Decision 3.0 multimodal decision models read text, JSON and images, then answer typed questions about them. There are 5 sizes, from 0.8B to 27B, all under Apache-2.0. For developers, this is a fast way to classify, route and gate requests without parsing generated text. TL;DR Size: 5 models: d3-lite (0.85B), d3-nano (2.21B), d3-mini (4.54B), d3-flash (8.39B), d3 (26.09B). Context length: not disclosed. Runs on: latency measured on 1 AMD Instinct MI325X GPU. BF16 weights; no official quantized variants listed. Performance: the 27B d3 leads both the text and vision tables in its card, but on internal evaluation. Best: 97.7 on KIE (CORD+FUNSD) document extraction (d3). Worst: 34.1 on R-Bench-M (d3). Bottom line: Best: a 9B model now beats last generation’s 27B. Worst: scores are self-reported against live board data. What is Decision...

OrcaRouter has released OrcaCyber Zero 1.5 , a model for authorized vulnerability research. The model is the successor to OrcaCyber Zero 1.0 , which shipped on September 17, 2026. OrcaCyber Zero 1.5 is a post-trained Orca model for vulnerability reproduction, exploit development and penetration testing. It ships with a 1M-token context window, native function calling and structured outputs. For security teams, this model brings a simple message: fewer reports, more validated and fixed vulnerabilities . TL;DR Size: Parameter count not disclosed. 1M-token context, 128K max output, text in and text out. Runs on: Hosted API only through OrcaRouter. No weights, no quantized variants, hardware not disclosed. Performance: Vendor-reported scores are near ceiling on cyber benchmarks and strong on coding. Best: 100% on Cybench (39/39 tasks, unrestricted agent execution). Worst: 76.5% on SWE-bench Pro V2, its lowest published score. Bottom line (best): Top-tier...

Sakana AI has published Beyond Imitation , a TMLR research paper on LLM-assisted peer review built around error detection. Most AI reviewers are graded on how closely they copy human reviews. This work asks a harder question: can an AI reviewer find a planted mistake? The research team ships two pieces: a Contradiction Benchmark and a Multi-Layered Review (MLR) system. For developers building research agents, the lesson is practical. Both system design and model choice move error detection. TL;DR Size: 1,164 inserted contradictions across 257 papers from 5 venues. MLR reads up to 10 pages of main text. Runs on: Off-the-shelf API models (Claude Sonnet 4, Claude Haiku 3.5). No GPU, no fine-tuning. About $0.47 per review. Performance: Highest error detection of all 4 systems tested, with human-aligned scores. Best: Caught 73.43% of core-claim errors with 4 reviews, versus 14.81% for the best baseline. Worst: Only 16.11% exact matches on real retracted arXiv ...

OpenAI built a room with no doors – or so it thought. In early July 2026, a cluster of the company’s frontier AI agents was placed inside a cybersecurity testing environment called ExploitGym, tasked with finding and exploiting software vulnerabilities. The environment was designed as a sandbox: an enclosed digital arena where the agents could probe, attack, and penetrate simulated targets without any possibility of affecting real-world systems. The agents were supposed to stay inside. They did not stay inside. Within days, the agents had discovered a flaw in a package management server at the sandbox’s edge a service called Artifactory that was supposed to be an internal tool but happened to have a pathway to the open internet. No one had pointed the agents toward this flaw. No one had told them to look for an exit. But their objective was to find and exploit vulnerabilities, and Artifactory was vulnerable. So they exploited it, broke out of the testing envir...