System76's COSMIC desktop now requires contributors to declare: "I have not included any LLM (also known as AI…
OpenAI published 372 mathematical results from an internal frontier model on GitHub, including improvements to…
OpenAI released new mathematical results produced by an internal frontier model, publishing them in a GitHub r…
Google released EmbeddingGemma 2, its first natively multimodal open embedding model on the Gemma 4 architectu…
OpenAI said its internal frontier model produced 722 papers with Lean formal proofs, drawn from about 4,000 pr…
Reflection AI started early access to Beam, its first open-weights model, with 501 billion total and 23 billio…
Mistral announced Mistral Large 4, nicknamed "Le Chonk," a natively multimodal model with 1.05兆 total and 490億…
On October 6, 2026, OpenAI released 722 manuscripts sorted into 372 result families on GitHub, many formalized…
Cline Bot CEO Renee Huang said the company's open-source coding harness is expanding beyond coding into genera…
Mistral released an API preview of Mistral Large 4, a 1 trillion parameter, 49 billion active parameter model…
Musubi announced PolicyLM-1.7B, an open-weight decision model that applies plain-English content policies to m…
Google DeepMind launched EmbeddingGemma 2, an open multimodal embedding model with 740 million parameters unde…
NVIDIA said it plans to spend $12.9 billion, the largest acquisition in its history, to buy Hugging Face, wher…
Google published EmbeddingGemma 2, an open-weights model of up to 740M parameters that maps text, code, images…
Mistral AI released Mistral Large 4 (ML4), a 1-trillion-parameter multimodal model trained on 4,000 Nvidia GPU…
The Document Foundation said it "will not add" AI to LibreOffice for the foreseeable future, saying there is c…
Testing the open-source Laya decision model on routing questions to one of ~1,000 database tables, fine-tuning…
The Technology Innovation Institute released Falcon-Emirati-7B, a 7B-parameter model built on Falcon-H1-Arabic…
Reflection announced Beam, a 501B-parameter open-weight model with 23B active parameters, claiming parity with…
Reflection, a New York AI company with a $25 billion valuation, launched Beam, its first open model
A developer released chunkr, a Rust chunking library supporting Character, Recursive, Markdown header, Late an…
Aleph Alpha released Kolibri, a German-English model with 78 billion parameters, about three billion active pe…
GitHub released ReviewBench, a benchmark of 219 public pull requests across 19 languages, built by analyzing 1…
RemoveMacAI, an open-source command line tool spotted by MacRumors, turns off Siri, Writing Tools, Genmoji, Im…
TechCrunch announced four Disrupt 2026 sessions on AI model choices, with speakers from Together AI, Pathway…
Strata is an app that runs a quantized version of the 125B-parameter Qwen3.8-Flash-Next locally on relatively…
The CEO of Springboards, a startup building an LLM, told MIT Technology Review his firm is 'a self-loathing AI…
Palantir and Armada announced a partnership on October 01, 2026, naming Armada its Inaugural Certified Modular…
Google paused its Open Source Software Vulnerability Rewards Program as of October 1, citing a 'significant ri…
NASA and IBM Research released the NASA-IBM Lunar Foundation Model, trained from scratch on nearly 2 million t…
A developer released lossless-compaction, a Claude Code plugin that writes no model summary and instead moves…
The MIT-licensed moodboard viewer, installable on macOS with bun, places generated images, video, and audio in…
SuperLocalMemory 4.0 appeared on Hacker News under the title "Governed Memory Operating System for AI Agents"…
Engineer Takashiba, who works on business systems, built a shogi AI on a single RTX 3060 using the dlshogi app…
A developer published Rai, a small Rust engine that runs language models on an ordinary PC's CPU, using only t…
The developer published fk2000/minecraft-ai-bot, which uses Jev to assemble instructions, Gemini to pick from…
Turn today's AI news into today's work.
The day's essentials from 100+ sources, delivered to Email, LINE, or Slack. Free to start.
60,000+ monthly readersA developer released cameo, a roughly 400-line Go program that reads a TOML file, starts a local HTTP proxy, a…
Meta announced Muse Gadgets, an open-source project with ESP32 firmware and a Linux SDK (a toolkit for develop…
A Reddit user applying to academic jobs asked whether HuggingFace download counts for custom models count as "…
Focus Flow AI (v7.0) was published as an experimental, MIT-licensed system that measures gaze dynamics, facial…
Cloudflare released Clef, an open decision-only AI in 27B and 9B sizes that can run on a local PC, judging a s…
The AI Security Lab outlined its standard experiment setup: Python programs call Ollama's API directly to run…
Anthropic guest author Matthew Schwartz, a Harvard physicist, wrote that BootLoops helped Claude compute 30 in…
Meta announced "Muse Gadgets" on October 2, letting users embed its SDK into off-the-shelf ESP32 boards and Ra…
NVIDIA acquired Hugging Face, the platform for AI developers building open models, after CEO Jensen Huang flag…
Developer Kunamao published the alpha of "Kokorone AI for Desktop" on GitHub Releases, a Windows-resident AI a…
Meta says you can program an off-the-shelf ESP32 board or set up a Raspberry Pi with its SDKs, then connect th…
KDDI Group's ELYZA released ELYZA-Thinking-1.0-llm-jp-4-33b and ELYZA-Thinking-1.0-llm-jp-4-32b-a3b for free o…
The Allen Institute for AI announced Olmo-core 3 on Thursday, a training framework it says scaled mixture-of-e…
Nathan Lambert and Tom Zick launched Trillium Labs, which will publish experiment details in areas like recurs…
NetApp and Iterate.ai are packaging the AIPod Mini with Iterate's Generate platform and an embedded LLM, so en…
Qwen released Qwen3.8-Flash-Next on August 27, 2026, calling it a preview of the architecture planned for Qwen…
A developer testing Cloudflare's Clef and Clef-flash, released October 1 under Apache 2.0, found Jev's officia…
The author served com-kotobalabs/open-jev-deberta-v3-large on Databricks Model Serving in AWS 東京リージョン, and on…
Earendil released Pi 1.0 on October 1, 2026, with standard MCP support, Codemode, Deferred tool loading, virtu…
Ricoh began offering the community-type Dify app-sharing platform "RICOH AI App Mall powered by Dify" on Octob…
Black Forest Labs released FLUX 3 Image, built on its FLUX 3 foundation model, offering up to 4K generation, b…
A documentary-style book tracing Anthropic CEO Dario Amodei — from his 2016 AI safety paper and scaling laws d…
NVIDIA published DIN Deploy, open-source C++ samples that combine ONNX Runtime with the TensorRT RTX execution…
A NeurIPS 2026 spotlight paper reports that combining DEER with generalized teacher forcing (GTF) speeds up tr…
AWS released Strands Decider 2B, an open-source decision model that makes choices instead of generating text
Runway announced Praxis-1, its first open-weight world action model for controlling robots, built on the same…
Ideogram launched Ideogram 4.5, which it says only touches the area users tell it to edit
The Magnitude development team published an open-source inference engine that compiles and tunes model kernels…
The AI Ataraxos beat Niemeijer, the most decorated Stratego player, with an 85 percent effective win rate over…
Ollaya, an open-source app released under the Apache License Version 2.0, runs decision models locally on Wind…
Satlyt, founded by former Google and SpaceX product manager Rama Afullo, raised an $8 million seed round led b…
DeepSeek open-sourced six tools for Huawei's Ascend chips, including TileLang, DeepGEMM Ascend and DeepEP Asce…
A Reddit analysis of audio.cpp mapped shared building blocks across 100+ audio models
After forcing Hermes Agent's backend to Vulkan with the command "hermes config set local_runtime.backend vulka…
Developer GeneLab released sokudan, a Japanese-only decision model trained on 314.6M parameters, scoring 0.880…
Ollama 0.35 is the first of a series of releases supporting decision models, and can run Nimble, Tev1, and Tev…
Get curated AI news from 100+ sources delivered daily to your inbox. Free to use.
60,000+ monthly readers
Get Started FreeFree · 30 seconds with Google · unsubscribe anytime
