NVIDIA PAIR, a free open-source Personal AI Router in open beta for Windows, macOS, and Linux, pools idle GPUs across home networks to route inference requests to available machines, including Apple M…
NVIDIA launched PAIR (Personal AI Router) on September 3rd, an open-source beta that routes parallel AI inference requests across Windows, Linux, and macOS devices on a local network, targeting multi-…
Samsung's Tiny Recursive Model (TRM), with only 7 million parameters and two layers, outperforms frontier models like GPT-4 on the ARC-AGI-2 benchmark, suggesting that architecture and recursive logic…
Nvidia's DLSS 5 technology, releasing this week with NBA 2K27, incurs a 50-60% performance drop when enabled, according to Nvidia, but the company has made it roughly five times faster than its origin…
NVIDIA confirmed DLSS 5 will launch on September 3 with NBA 2K27 as its only supported game, and the neural renderer still costs 50-60% of native frame rate before Multi Frame Generation. The company …
The State of Local AI Report, updated monthly, maps what AI models can run on user-owned hardware, finding that memory bandwidth dictates decode speed while compute dictates prefill speed, with an RTX…
NVIDIA's DLSS 5, its latest AI upscaling technology, launched Sept. 3 exclusively for RTX 50-series graphics cards with only one game, NBA 2K27, and has sparked controversy over AI redrawing game visu…
Nvidia's DLSS 5 launches September 3 at 9 p.m. Pacific with NBA 2K27, and will only officially support GeForce RTX 50 series GPUs and GeForce NOW Ultimate cloud rigs, excluding RTX 40 and older cards.…
NVIDIA announced that its DLSS 5 neural rendering model is now five times faster than when first shown at GTC in March, requiring only a single RTX 50-series GPU instead of two RTX 5090s. The feature …
Nvidia's DLSS 5, launching September 3rd alongside NBA 2K27, now runs on a single RTX 5090 instead of two, a 5x performance improvement since March, but the GPU's street price has risen to at least $5…
Nvidia's DLSS 5 launches September 3rd with NBA 2K27, requiring RTX 50-series GPUs or GeForce Now, and demands serious hardware: even a midrange RTX 5060 needs a top CPU and 6x frame generation for 10…
Modders have enabled Nvidia's unreleased DLSS 5 AI upscaling technology on older RTX 40 and RTX 30 series graphics cards, including an RTX 3080, but performance is severely degraded, with even the RTX…
A new arXiv paper (2608.28003v1) proposes a layer bit-width allocation method for Gemma-3-1B that maximizes latency reduction under a quality-degradation constraint, achieving an 11.0% latency reducti…
A developer has launched a self-hosted API on Base mainnet that sells LLM inference for USDC via the x402 HTTP 402 payment protocol, with a live endpoint running on a single RTX 5090 GPU and pricing s…
A developer reported that LM Studio's runtime auto-update silently broke model loading on an RTX 5090, with the error dialog showing an unsigned exit code that decodes to an NTSTATUS value indicating …
A developer implemented a generative AI image model on a $1 RP2350 microcontroller, as used in the Raspberry Pi Pico 2, generating 128×128 RGB images of human faces in 10–20 seconds each. The model, a…
A Hacker News user proposed an OpenAI-compatible API for open-weight models that guarantees zero prompt retention, no training on user data, and logs only billing metadata. The user, who is considerin…
Nvidia released DLSS 4.5 Ray Reconstruction on August 25, 2025, with support for 30 games and compatibility with all GeForce RTX GPUs from the RTX 20 series to the RTX 50 series. The update introduces…
Escha Labs released Qwen3.8-27B-Escha-W2, a 2-bit quantized build of Qwen3.8-27B that fits 10.15 GB of weights on a 24 GB consumer GPU, enabling 64k context out of the box and up to 128k with tuning v…
A user running Qwen3.8-27B on an RTX 5090 with the ninfer runtime reported a 1.72× speedup over LM Studio/llama.cpp, achieving 124.08 tok/s decode and 8,143.82 tok/s prompt eval with mixed NVFP4/FP8 q…