AI Pulse by Inblix
Sunday, September 20, 2026 AutoCurated · Updated 11:15 UTC

AI Research

Cutting-edge AI research papers, breakthroughs, and academic developments.

1153 articles · page 30

Latest

Synchronous batching wastes 24% of GPU time: Here's how async cuts it to zero

May 14

GPU Math: Why AWS's Latest Instances Won't Fix Your Memory Bottleneck

May 11

Four V1 bugs nearly sank our RL training — here’s what vLLM 0.18.1 actually fixed

May 6

Hugging Face locks down new speech datasets to stop AI benchmark gaming

May 6

IBM’s Granite 4.1 packs 30B dense model punch using just 15T tokens and a 5-stage data diet

Apr 29

DeepInfra joins Hugging Face Hub, slashing serverless AI costs for devs

Apr 29

NVIDIA's New 30B Model Is 9x Faster at Parsing Your 100-Page PDFs and Video Archives

Apr 28

OpenAI's privacy filter runs locally with just 50M active parameters, enabling 128k-token document scans

Apr 27

DeepSeek's V4 slashes KV cache to 2% of normal size, making million-token agents practical

Apr 24

Running Local AI in a Chrome Extension: The MV3 Architecture That Actually Works

Apr 23

Up to 40% of Arabic AI Benchmarks Are Broken, QIMMA Audit Finds

Apr 21

Mythos cracked vulns fast — but its real lesson is the system, not the model

Apr 21

AI code agents flood open source with 10x more pull requests no one can review

Apr 16

Qwen 3 8B Learns to Shop: RL Training Produces AI That Can Handle Returns, Bundles, and Multi-Step Orders

Apr 16

Tom Aarsen's finetuned 2B model beats giants 4x its size on visual doc search

Apr 16

VAKRA Benchmark Exposes a Brutal Reality: Top AI Agents Fail at Basic 3-Step Workflows

Apr 15

HCompany's HoloTab lets anyone automate browser tasks with one recording

Apr 15

Overworld's Waypoint-1.5 Runs Real-Time AI Worlds on Your Own GPU

Apr 9

Qwen's New Multimodal Models Let You Search Images With Text—No Labels Needed

Apr 9

Hugging Face hands Safetensors to Linux Foundation to keep model format truly open

Apr 8

Google's Gemma 4 packs GPT-rivaling power into a 4B-parameter model

Apr 2

A 0.6B model just beat SAM 3 on grounding—here's the simple trick that made it work

Apr 1

Gradio's New Server Class Lets You Build Full-Stack AI Apps in 50 Lines of Python

Apr 1

IBM's 3B Granite Vision model beats much larger rivals at reading charts

Mar 31