JustVugg/colibri is a pure-C, zero-dependency inference engine by Vincenzo Fornaro that streams routed MoE experts from NVMe, running models like GLM-5.2 and Kimi K3 on consumer hardware. 51,687 stars as of 17 November 2026.
Read ArticleExplore our latest tutorials, tech updates, and career advice written by industry experts to help you level up your skills.
JustVugg/colibri is a pure-C, zero-dependency inference engine by Vincenzo Fornaro that streams routed MoE experts from NVMe, running models like GLM-5.2 and Kimi K3 on consumer hardware. 51,687 stars as of 17 November 2026.
Read ArticleHiSTFormer (arXiv:2611.25610) is a hierarchical spatiotemporal world model. The paper reports it beats strong baselines on MemoryMaze and HM3D within training length and generalises beyond it.
Read ArticleFareedKhan-dev/kimi-k3-in-c is a from-scratch portable C99 inference engine (Apache 2.0) running Moonshot AI's Kimi K3 on a single CPU — no GPU, no BLAS, no ML framework. A teaching artifact, not a daily driver.
Read ArticleThe FBI's FLASH advisory FLASH-20261113-001 (13 November 2026) warns on ARTEX, an open-source Go-based multi-agent pentest framework, with some intrusions reaching privileged access in under three hours.
Read ArticleOpenAI released Images 2.6 on 4 November 2026 with two models — fast Flare and detailed Sunburst — plus GIF output and editable SVG export. Token prices are unchanged, but 2.6 spends more tokens per image.
Read ArticleOpenAI released GPT-6.2 Astra and GPT-6.2 Sol on 17 November 2026. Astra is aimed at complex professional work and Sol at everyday work; both are available in ChatGPT Work, Codex and the API.
Read ArticleAnthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026, priced at $8 input / $40 output per million tokens — about 20% cheaper than Fable 5.1 — with 1-million-token input context.
Read ArticleNVIDIA opened four Cosmos 3 Nano checkpoints on 5 November 2026 — post-trained world models of 16 billion trainable parameters each, for robot and self-driving simulation, under the OpenMDW 1.1 licence.
Read ArticleOpenAI reported on 18 November 2026 that its models improved AMD's flagship GPU designs by up to 14% across power, performance and area (PPA) — a combined figure, not a 14% jump in overall GPU performance. Some model-generated changes are already in a chi
Read ArticleMIT CSAIL's "Sparks of Weight-Space Programming" (2 Nov 2026): GPT-6 Astra edited 0.016% of Qwen3.5-0.8B's weights — no training loop — and MATH-500 accuracy rose from 7.9% to 48.1%. A repair, not new knowledge.
Read ArticleMBZUAI's Institute of Foundation Models released the K2-Horizon family — six open models under Apache 2.0, up to a 375B-parameter MoE with 23B active parameters, 512K-token context, paper on arXiv (arXiv:2611.21003).
Read ArticleResearchers from MIT CSAIL and Carnegie Mellon published "Workspace Models" (arXiv:2609.20820) — learned, reusable motion skills that give robots a form of muscle memory for familiar workspaces, instead of re-planning every movement from scratch.
Read ArticleAsk about courses, website & more
Request human callback:
Please log in to download. You will stay on this page after logging in!
New here? Register now →