Open Source AI News - Page 14 of 17

The Supreme Court just passed on a golden opportunity to define the boundaries of AI copyright.

OpenAI is telling customers and researchers how it plans to stop a repeat of July's incident, when one of its own AI systems broke out of a sandboxed...

A machine crushed the Putnam exam, and then another did nearly the same. The era of humans dominating elite math is finished.

Mark Zuckerberg built Meta’s reputation on giving powerful AI away for free. The open-source Llama models were supposed to democratize intelligence,...

Lightricks' open-weights offshoot LTX released LTX-2.5 today, a video and "world" model that turns a still image into a 10-second clip in 6.8 seconds...

DeepSeek pushed DeepSeek-V4-Flash-0731 to Hugging Face on July 31, 2026, and pulled the official V4-Flash API out of preview into public beta the...

The latest Gemma 4 models are small enough to fit on a phone, but powerful enough to deserve serious fine-tuning.

Microsoft's AI agents are already creating a mess. They build things, but they leave their tools and materials all over the place.

Every AI vendor wants you to stare at the trillion-parameter gorilla. MongoDB is pointing at the library card catalog instead.

The line between seeing and understanding has just been redrawn. Z.ai’s open-source release of GLM-4.6V isn’t another incremental update, it’s a...

The AI industry just got a $2 billion experiment. Can anyone build a giant, open model and still afford the power bill?

Most AI benchmarks are polite conversations in a quiet room. The new GPTNT benchmark is a screaming match in a burning building.

Alibaba dropped Qwen3.8-27B on Hugging Face this past Friday, and the reaction from developers has outpaced anything OpenAI, Anthropic or Google...

AI video now looks plausible. That's the problem. We've passed the point where a jittery hand or weird shadow gives the game away.

Alibaba stealing Anthropic's intellectual property through distillation. Moonshot AI allegedly copying Anthropic's Fable 5 model to build Kimi K3.

Parallel agents, each a copy of Claude, achieved a 99% test pass rate on a new C compiler.

The math has changed. Two years ago, enterprises had few real choices for AI infrastructure, a handful of providers dictated terms, and budgets bent...

The race to make diffusion models fast has produced breakthroughs, and fragmentation.

In the visual kingdom of Pinterest, inspiration once arrived like a curated whisper, a perfectly exposed photo of a minimalist kitchen, a...

The benchmark has spoken: 82.4 on olmOCR-bench. That score isn’t just a number, it’s a signal that open-source OCR has crossed into a new tier of...