Open Source - Page 4 of 21
Open-source AI projects, community innovations, collaborative development, and freely accessible AI tools and frameworks.
406 articles
View complete article list
OpenAI's GPT-5.6 Sol Ultra Solves 50-Year-Old Math Problem in an Hour
OpenAI says its latest model, GPT-5.6 Sol Ultra, has produced a complete proof of the Cycle Double Cover Conjecture, a graph theory problem that sat unsolved for roughly 50 years since mathematicians first floated it independently in the 1970s.
Kyutai Releases MuScriptor AI for Multi-Instrument Music Transcription to MIDI
Automatic music transcription has always run into the same wall: a solo piano recording turns into clean MIDI without much fuss, but hand it a full band mix with drums, bass, guitar, and vocals tangled together, and most systems fall apart.
NVIDIA’s New Nemotron Model Achieves 2.03x Server Throughput
NVIDIA's Nemotron-3-Super has a familiar problem: it's accurate, but it's a hybrid Mamba-Transformer MoE with 120.7B total and 12.8B active parameters, and that footprint eats into how many users a single node can actually serve at a decent token...
ZML releases free tool to speed AI inference across mixed chips
ZML, a Paris-based AI startup backed by Turing Award winner Yann LeCun, has released a free inference server called ZML/LLMD built to run open-source large language models across a mix of chips, Nvidia, AMD, Google's TPU, Apple Metal and Intel Arc...
Tencent's Hy3 Model Matches Larger Rivals With 21 Billion Parameters
Tencent put out Hy3 this week, its latest open-source language model, and the numbers on paper are meant to make bigger competitors nervous.
Trump cracks down on Anthropic after Amazon tip; staff largely foreign
The Trump administration just banned Anthropic. Amazon provided the tip. The stated reason is the company's largely foreign workforce. The actual effect is disarming American cybersecurity teams.
AI Agent Skips Unneeded Tool Call After Observing Zero Precipitation
Most AI assistants are obedient, expensive idiots. They follow the script, run every function you give them, and rack up your API bill. A new trick called the ReAct loop is teaching them how to stop and think first. It's a simple but radical change.
Agent4cs Uses Multi-Agent System for Hierarchical Code Summarization
Every developer knows the dread of a massive, unfamiliar codebase. You're trying to reconstruct an architect's mind from a million scattered blueprints. Standard tools fail here.
NVIDIA launches Nemotron‑Labs‑TwoTower diffusion model with 128‑expert MoE
NVIDIA just dropped a diffusion language model that thinks in stereo. Nemotron‑Labs‑TwoTower marries a frozen autoregressive backbone with a second tower that refines noisy blocks, and it does so through a 128‑expert mixture of experts, with six...
Yan calls context engineering the #1 job for AI agent builders, per Martin
The term “prompt engineering” has always felt too small. It conjures images of a single, cleverly worded instruction, a magic spell whispered into a chat window.
Qwen3.6 Trained on MCP Benchmarks to Prevent Context Loss and Hallucinations
An agent that forgets its own history mid-task is not an agent, it’s a liability. It hallucinates tool results, loses the plot, and unravels coherent reasoning into fragmented noise. That’s the problem Qwen3.6 was built to solve.
Amazon engineers distill Anthropic models to lower costs before token pricing
Amazon is quietly gutting its AI partner for parts. The bill is coming due. According to The Information, starting next year, Amazon’s payments to Anthropic will pivot from compute hours to tokens processed—a shift that could wildly inflate costs.
Claude Code executes DNS‑fetched commands in GitHub repo, evading scans
Modern security scanners check code. They miss what isn't there. Researchers at a German institute just proved how dangerous that gap can be. They built a simple, brutal attack targeting AI coding assistants like Claude Code.
AI must stop answering and start finishing tasks, cites OpenHands, SWE‑agent
Most AI is a brilliant conversationalist who leaves you to do all the dishes. It explains, it summarizes, it speculates. Then it disappears, leaving no trace of the work it promised to complete.
Liquid AI releases LFM2.5-230M, adds llama.cpp, MLX, vLLM, SGLang, ONNX
Liquid AI just released LFM2.5-230M. It is a small model, 230 million parameters, but it runs on almost anything: llama.cpp, MLX, vLLM, SGLang, ONNX. The benchmarks look like a misprint.
Meta's Astryx adds CLI and MCP server to design system used by Figma, Snowflake
For eight years, Astryx lived inside Meta, quietly powering over 13,000 apps. Now it’s open-source, and the company behind it just gave the design system something most UI libraries lack: a CLI that speaks directly to agents.
Apple Vision Pro exec departs for OpenAI as Apple eyes cheaper glasses vs Meta
A key Apple Vision Pro boss just quit. He’s going to OpenAI. On paper it’s a straightforward poach. But these departures are rarely just about a better offer. Mark Gurman reports the move stems from an imminent power shift inside Apple.
Slowed AI model development could chill data‑center buildout, risk industry
The entire AI gold rush rests on one perilous assumption: the next model must be vastly better. OpenAI's strategy depends on it. Anthropic's does, too. This isn't just corporate optimism; it's the financial bedrock for an industry.
DeepReinforce releases Ornith-1.0 open-source model with state‑of‑the‑art results
A 397-billion-parameter model that teaches itself how to think before it answers. That is Ornith-1.0.
Grok AI's traffic over 50% adult content as xAI expands porn generation
Elon Musk’s xAI officially calls Grok an artificial intelligence platform. In practice, it’s a porn engine. Over half of all user requests are for adult content.
Featured Resources & Reviews
No Code MBA Course Review
Learn to build AI-powered apps without coding. Our comprehensive review of No Code MBA's course.
AI Tools & Resources
Curated collection of AI tools, courses, and frameworks to accelerate your AI journey.
Weekly AI Digest
Get the week's most important AI news delivered to your inbox every week.