OpenAI GPT-5.5 Codex Bug Clusters Reasoning Tokens and Degrades Performance
A newly uncovered bug in OpenAI’s GPT-5.5 Codex causes reasoning outputs to cluster at rigid boundaries, resulting in degraded code generation.
A newly uncovered bug in OpenAI’s GPT-5.5 Codex causes reasoning outputs to cluster at rigid boundaries, resulting in degraded code generation.
Fable has launched Splat4D, an open-source, streamable 4D Gaussian splatting format that slashes raw file sizes by up to 58 times for web-native 3D scenes.
A new open-source project brings Command & Conquer Generals natively to macOS, iPhone, and iPad using the original engine and modern renderers.
A massive surge in high- and critical-severity cyber vulnerability disclosures followed the internal deployment of Anthropic’s autonomous security model.
A new local proxy tool called pxpipe reduces Claude Code token usage by converting bulky text and code context into images before sending them to the model.
Giving AI agents search access to previous session transcripts fails to improve coding performance and can actually degrade output quality.
A new behind-the-scenes video of Midjourney’s hardware-based medical scanner offers a closer look at the machinery but lacks scientific proof of its efficacy.
A security-focused researcher has introduced the ‘Short Leash’ AI coding method to prevent autonomous agents from generating inefficient code or losing codebase context.
A new open-source utility called claude-real-video processes video files locally, extracting scene-aware frames and transcripts for any large language model.