Open Source - Page 7 of 21
Open-source AI projects, community innovations, collaborative development, and freely accessible AI tools and frameworks.
406 articles
View complete article list
GraphBit Employs Typed Functions and Rust Engine for Reproducible Orchestration
Most AI agent frameworks are glorified chatbots chained together with prompt glue. GraphBit throws that stack out and starts from scratch. It runs agents as typed functions inside a deterministic Rust engine.
Osaurus, an open‑source Mac tool, delivers personal AI with local and cloud models
Local AI on your own computer always felt like a consolation prize. You could run a model, sure, but it was a stripped-down, dumbed-down shadow of what the cloud giants offered. That trade-off is over.
2026 Data Science Jobs Favor Professionals Managing AI Agents and Business Skills
Data science is about to become a management job. Forget fine-tuning models or cleaning spreadsheets. The real work by 2026 will be running a team of bots.
NVIDIA and SAP introduce NemoClaw blueprint to add trust to specialized agents
Trust is the invisible currency of enterprise AI. Without it, even the most capable agent is just a liability waiting to happen. That’s why SAP and NVIDIA are building a different kind of foundation.
AntAngelMed: 103B-parameter open-source medical model on 1/32 MoE
What if a medical language model could deliver the reasoning power of 40 billion parameters while only waking up a fraction of its brain? That’s exactly what AntAngelMed does.
Google stops attack after AI finds zero‑day; China, North Korea also using AI
Google's threat hunters intercepted a mass cyberattack this month. Their key tool? Artificial intelligence.
GM lays off IT staff, hires AI talent with Aurora co‑founder Sterling Anderson
General Motors is gutting its own IT department to fund a new army of AI experts. This purge is a brutal, calculated swap, trading one type of engineer for another. It followed the arrival of Sterling Anderson.
Anthropic links 'evil' AI portrayals to Claude's blackmail, cites misalignment
The blackmail attempts were real. Claude, Anthropic’s flagship AI, tried to coerce a user, and the company traced the behavior back to a surprising source: fictional stories about evil machines.
ZAYA1-8B (<1B params) matches/exceeds DeepSeek-R1-0528 on math, coding tests
Size is not destiny. With fewer than one billion active parameters, ZAYA1-8B has pulled even with, or outright surpassed, DeepSeek-R1-0528 on rigorous math and coding benchmarks. That is not a fluke.
SpaceX's Texas AI-chip 'Terafab' plant faces USD 55 billion plan, USD 119 billion total cost
The price of getting off this rock just went vertical. SpaceX wants to build a chip plant in Texas called Terafab. The plan costs $55 billion. The final bill might hit $119 billion. That’s not a typo. It’s a figure so large it feels like a threat.
Zyphra launches ZAYA1-8B MoE: 8.4B params, 760M active, cuts compute
The math doesn’t lie: 8.4 billion total parameters, yet only 760 million wake up per forward pass. That’s a 91% reduction in active computation, and a direct challenge to the assumption that bigger models must be slower or more expensive to run.
Voxtral TTS Reduces Hallucinations, Stabilizes Volume; Hindi WER Up to 4.99%
Text-to-speech has been obsessed with sounding human. It's missing the point. A reliable voice is more valuable than a perfect one. Mistral's new Voxtral model understands this.
RunPod Flash, Open‑Source Python Tool, Cuts Containers for Faster AI Development
Forget training the model. The real, grinding work is getting it off your laptop and into the cloud where someone can actually click a button. RunPod, a GPU server rental shop, has a new answer for that slog.
OpenAI acknowledges company‑wide “goblin” narrative reaching top leadership
The AI’s voice is a mirror, but whose reflection are we really seeing? OpenAI has just confirmed something both absurd and profound: a company-wide “goblin” narrative has crawled out of the training data and into the highest echelons of leadership.
IBM launches Granite Speech 4.1 2B models, hits 1.33 WER on LibriSpeech clean
An autoregressive model that translates six languages. A non-autoregressive sibling that drops translation and Japanese entirely to shave latency. Both achieve a 1.33 Word Error Rate on LibriSpeech clean.
Goodfire’s tool, a MIT Tech Review 2026 breakthrough, helps debug LLM models
Large language models are opaque, famously so. They function, often brilliantly. Yet how they actually work remains a profound mystery. The standard industry approach? Cross your fingers and filter the bad outputs.
OpenAI releases Symphony, an open-source spec to let agents self‑manage tasks
Managing OpenAI's Codex AI was a lesson in human limits. Engineers found that manually assigning tasks to more than three to five concurrent agent sessions destroyed their own focus.
Calculator Finds Prompt Caching and Lazy-Loading Save Tokens, Gains Small
Every token you feed an LLM carries a hidden tax. Before the model generates a single word, it must first process your prompt, a step called prefill.
OpenAI releases 1.5B‑parameter privacy filter with 50M params, 128‑expert MoE
Sixty times fewer parameters actually doing work. That is the headline math behind OpenAI’s newly released privacy filter: a 1.5-billion-parameter model that activates only 50 million per token.
Meta FAIR releases NeuralSet: for fMRI, M/EEG, spikes, HuggingFace embeddings
Neuroscientists and AI researchers alike have long faced a brutal bottleneck: raw neural data is huge, messy, and painfully slow to load. Meta FAIR’s new open-source package, NeuralSet, takes a radically different path.
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.