Colibri Runs a 744B Model on a Laptop With No GPU
A single-file C engine streams experts from disk to run 744B-parameter MoE models on a 25GB laptop, no GPU required.
// Intelligence
Today in AI
A focused feed of models, agents, research and open-source releases for people building with AI.
Updated
Grok App Turns xAI's Terminal Agent Into a Full Desktop Workbench
An unofficial Tauri 2 desktop app wraps xAI's Grok Build CLI with sessions, project management, media previews, and scheduled automations.
Hugging Face's Tau Teaches Developers How Coding Agents Actually Work
Hugging Face released Tau, an MIT-licensed terminal coding agent designed as much for reading as running, with a three-layer architecture you can actually understand.
Sub2API Lets Teams Share Claude and OpenAI Subscriptions Like Carpooling
A Go-based open source relay platform pools Claude, OpenAI, Gemini, and Grok subscriptions behind one endpoint with billing, sticky sessions, and cost sharing.
DeepSeek's V4-Flash-Vision-Exp Quietly Challenges Anthropic's Opus on Multimodal Agent Tasks
DeepSeek's new experimental vision model bolts image understanding onto V4-Flash at the same price, closing in on Claude Opus 4.8 on multimodal agent tasks.
Stanford's Marin Trains a 535B Open Model Anyone Can Watch Live
Stanford's Marin lab kicked off a 535B mixture-of-experts run with 23B active parameters, streaming every metric, config, and mistake live to the public.
LY Corporation's sim-use Gives AI Agents Eyes and Hands on Mobile Apps
A Swift CLI from LY Corporation gives AI coding agents a token-efficient observe-act loop across iOS Simulator and Android devices.
Kizuna AI Lab's Sokuji Translates Live Calls Offline Without a GPU
Sokuji is an open-source live speech translator that runs entirely on-device with WASM and WebGPU, no API key or GPU required.
Magnitude Turns Local AI Model Selection Into a One-Click Decision
Magnitude's new catalog profiles your hardware, estimates tokens per second for every model, and picks the best local LLM before you download anything.
Sakana AI's Namazu Beats Google Translate and DeepL on Japanese Cultural Nuance
Sakana AI upgraded its free JP-EN-ZH translator to the new Namazu model, beating Google Translate, DeepL, and Claude Opus 4.8 in head-to-head evaluations.
AEON vLLM Ultimate Fixes NVIDIA DGX Spark's Broken AI Stack in One Pull
A community-built vLLM container brings NVFP4 KV cache, DFlash speculative decoding, and Blackwell sm_121a runtime patches to DGX Spark serving.
Alibaba's Qwen-Image-3.0-Pro Jumps 83 Elo Points but Drops Open Weights
Alibaba's third-generation image model debuts at #6 in editing and #9 in text-to-image, with big Elo jumps and a productivity-first pitch.
Anthropic Ships Claude's Agent Stack With 40% Fewer Round Trips
Anthropic makes computer use, browser tool, Skills API, and Files API generally available with batched actions cutting round trips 20-40%.
OpenAI's GPT-Image-2 Now Generates Transparent Backgrounds in a Single API Call
GPT-Image-2 can now generate PNGs with real alpha channels directly through the API, skipping the separate background-removal step for cutouts.
Kaggle's Adversarial Customer Service Benchmark Pits AI Against Identity Thieves
Kaggle and Gert Labs turned identity-theft prevention into a two-model roleplay, testing whether LLMs can catch social engineers without stonewalling real customers.
Qwen3.8-27B Unleashed Hits 83% MMLU Uncensored on a Single 24GB GPU
A per-tensor dynamic quantization of an abliterated Qwen3.8-27B lands on Hugging Face, delivering 82.98% MMLU and 262k context on a single RTX 4090.
Soup Trains Llama-3.1-8B on a 4 GB Laptop GPU for Free
Layer streaming plus 4-bit quantization pins an 8B model into 3.32 GB of VRAM, turning laptop GPUs into real fine-tuning boxes.
Meta's WildArtifactBench Ditches Fixed Rubrics to Judge Real-World Agents
Meta previews WildArtifactBench, an evaluation that judges agents on messy real-world tasks using human and AI preference votes instead of rigid rubrics.
Google's Gemini 3.7 Flash Hits 84.6% on AGI Benchmark at 2.4x Lower Cost
Google's mid-tier reasoning model nearly clears the 85% target on ARC-AGI-2 at a quarter per task, redrawing the cost-performance frontier.
Google's Antigravity Brings AI Agents Into VS Code, JetBrains, and Zed
Google's agent-first coding platform now embeds directly in VS Code, JetBrains, Zed, and Visual Studio, sharing one account and context across editors.
Liquid AI's DSpark Triples LFM2.5 Speed Without Changing Its Outputs
Liquid AI ships DSpark draft models for its LFM2.5 family, delivering up to 3.18x GPU throughput and cutting agent latency by nearly half.
AT&T Bets on Hark to Fix What Killed Humane's AI Pin
AT&T takes an equity stake in Brett Adcock's stealthy AI hardware startup Hark, providing cellular connectivity for phone-free AI devices.
Meta Reveals Muse Spark 1.2's Multimodal Jump From 59.8 to 72.0 With Tools
Meta shared new benchmarks, robotics demos, and a real-world agent evaluation for Muse Spark 1.2, its coding-focused multimodal model ahead of an open-weights release.
Exa Plugs 100B Websites Directly Into OpenAI's Codex Agent
Exa's search API now plugs directly into Codex and ChatGPT, giving OpenAI's coding agent live access to over 100 billion web sources.
No stories match these filters.