Tag
A missing .npmignore entry exposed Claude Code's full TypeScript codebase, revealing hidden features, anti-distillation tricks, and an unreleased always-on daemon called Kairos.
Eight labs unite under NVIDIA's Nemotron Coalition, LangChain open-sources the enterprise coding agent pattern, and Sarvam proves frontier AI doesn't require Silicon Valley.
An 82-million parameter model that runs on a CPU, sounds nearly as good as ElevenLabs, and costs nothing. Here's how to set it up.
The fastest-growing open source project in GitHub history has become 2026's first major AI security disaster, with 135,000+ exposed instances, 9 CVEs in 4 days, and malware-laced skills.
Mistral Small 4's 119B MoE unifies reasoning, vision, and coding—but needs datacenter hardware. Qwen 3.5 35B-A3B remains the consumer GPU king at 112 t/s.
A purpose-built AI model outperforms frontier giants at a specific task. Is this the future of AI competition?
We compare four AI code review tools on real bugs, false positive rates, and value. The results show where AI excels and where it still falls short.
Hugging Face's Spring 2026 report reveals China now leads in AI model downloads, robotics datasets jumped 2,200%, and open-weight models are achieving 10x-1000x cost advantages.
Mistral releases a 4B parameter text-to-speech model that clones voices from 3 seconds of audio, runs locally on 16GB GPUs, and beats ElevenLabs in human evaluations.
Benchmark comparison of open-source OCR tools you can run locally. Surya, PaddleOCR, OlmOCR-2, and Tesseract tested on real documents.
New AI model predicts neural responses to any stimulus without scanning real brains, potentially accelerating neurological disease research
MiMo-V2-Flash runs 309B parameters on RTX 4090s. GLM-5 sets benchmarks but needs datacenters. Llama 4 Scout stays out of reach.
UC San Diego researchers built an AI that translates plain English questions about weather data into code and answers, aiming to democratize climate research
New compression algorithm achieves 6x memory reduction with zero accuracy loss. No retraining required. This matters for anyone running local AI.
Los Alamos researchers crack the configurational integral using tensor networks, making materials calculations that took supercomputer hours finish in seconds
UC San Diego researchers build an AI agent that translates natural language queries into climate model analysis, presenting at ICLR 2026
The creator of Gitleaks releases a faster, more accurate successor with 98.6% recall and native AI agent integration. Here's why it matters.
NVIDIA's Nemotron 3 Super runs agents locally, OpenAI releases Apache 2.0 models for the first time since GPT-2, and Alibaba's 9B parameter model outperforms 120B competitors.
The $29 billion code editor's new AI model outperforms Opus 4.6 at 1/10th the cost. Then developers discovered it's Kimi K2.5 from Beijing—and Cursor never told them.
Qwen 3.5's MoE models hit S-tier benchmarks, NVIDIA's Nemotron 3 Super delivers 5x throughput gains, and GLM-4.7-Flash brings frontier coding to consumer GPUs. The open-weight race just accelerated.
An open-source AI agent using interactive scaling beats OpenAI's GPT-5-high on Humanity's Last Exam. Here's what makes it different.
This week's open-source highlights: GPT-OSS marks OpenAI's first open weights since GPT-2, Superpowers becomes the most-starred AI coding framework, and Hunter Alpha was Xiaomi all along.
Jensen Huang bets on inference chips, Ollama adds multimodal support, and DeepSeek V4 remains the most anticipated release that hasn't happened yet.
Mistral drops a 119B MoE model under Apache 2.0, DeepSeek V4 emerges from stealth, and dual RTX 5090 setups are matching H100 on 70B inference. This week changed the game.
