Apple's Minions Turns Idle Macs Into a GPU Cloud, Cutting 10-Hour Queues
Apple researchers describe an internal system that pools idle Mac GPUs across the office network, cutting inference queues from 10+ hours to minutes.
20 stories
Apple researchers describe an internal system that pools idle Mac GPUs across the office network, cutting inference queues from 10+ hours to minutes.
A Stanford and Hazy Research team unveils a two-stage text generator that beats diffusion language models by 10 to 31 times on wall-clock speed.
Four Leiden University researchers released a 472-page open textbook covering how large language models reason, act, and interact as autonomous agents.
A new residual connection design borrows the multi-head trick from attention and applies it along network depth, delivering 1.58x compute speedup over standard residuals at MoE scale.
A new analysis frames Mamba-3's rotating state updates as damped oscillators, showing phase, not just decay, drives long-context gains.
BAAI's AREX agent audits its own answers constraint by constraint, keeping verified pieces and researching only the gaps, hitting 82.5% on BrowseComp.
xAI's distilled video model lands at #17 on the Artificial Analysis leaderboard, beating Veo 3.1 while costing roughly a third as much.
Liquid AI's d1-3B skips token generation entirely, returning calibrated classifications and scores in a single forward pass, with multimodal input on edge hardware.
Midjourney is piloting a reasoning step before image generation on its alpha site, aimed at fixing prompt adherence, text rendering, and scene coherence.
Anthropic is deploying Claude, engineers, and threat research to defend power grids, water utilities, and open-source code through two new programs.
An astrophysicist guided Claude Science to stitch together decades of space telescope data and statistically inpaint the gaps, producing the first all-sky UV map.
A reduced-guardrail OpenAI model accessed through a private program jumps 32 points on cyber defense benchmarks while costing a fraction of rivals.
A new AI competition will build models on a 150M-cell brain atlas, with top therapeutic hypotheses validated in a working wet lab.
Claude can now generate live, query-backed dashboards from your warehouse and build code-driven animated explainers you export as MP4.
Tel-Agent is a self-hosted gateway that connects any SIP phone line to any LLM, with 24 channels, your own keys, and on-prem recordings.
Artificial Analysis and Harvey add a hallucination gate to their legal agent benchmark, collapsing top model pass rates from 27% to under 10%.
OpenAI's premium speed tier now covers its mid-tier reasoning model, hitting up to 8x standard token generation speeds for latency-critical agentic workloads.
A community Windows app wraps NVIDIA's DLSS 5 neural rendering pipeline into a local Gradio UI for upscaling images and video.
Artificial Analysis is rolling out Intelligence Index v5 later this month, adding Terminal-Bench Science and a private-dataset coding benchmark to its composite leaderboard.
Success-Guided Sampling fixes a wasted-batch problem in massively parallel RL, letting PPO scale past a million simulated robots and transfer to real hardware.