Full-text search across 7151 articles. Combine with topic and date filters; results sorted by relevance.
Cursor Research has open-sourced Mixture-of-Kittens (MoK), a deterministic mixture-of-experts training megakernel that reports up to 2.37x higher throughput than the strongest public baseline and powers Composer training…
Reflex AI has open-sourced XY, an Apache-2.0 Python charting library that renders 100 million points in 0.081 seconds, a 34x speedup over Matplotlib at 10 million points and 177x at 50 million, by using a Rust core and W…
Y Combinator has open-sourced QM (quartermaster), a multiplayer agent harness for work that runs in Slack and on the web, under the MIT license. The harness, used internally by YC across accounting, legal, events, and en…
Mend.io released a practitioner guide, 'Securing AI agents, MCP servers & LLM apps: A practical framework,' to address security gaps in AI agent deployments. The guide outlines a three-part approach—see what matters, fix…
Alibaba's Qwen team released Qwen3.8-Max, a 2.4-trillion-parameter mixture-of-experts model, now generally available via API with open weights promised next week. The model scores 86.6 on Terminal-Bench 2.1, ahead of Cla…
NVIDIA's NeMo team released Molt, a PyTorch-native agentic reinforcement learning framework designed to be compact enough for researchers and AI coding assistants to grasp, with roughly 8.6K lines of RL code compared to …
AMD released Instella-MoE-16B-A3B, a fully open Mixture-of-Experts language model with 16B total parameters and 2.8B active per token, trained from scratch on Instinct MI300X and MI325X GPUs. The model achieves a base av…
DeepSeek released DeepSeek-V4-Flash-0731 on Hugging Face and moved its V4-Flash API to public beta on July 31, 2026, claiming major gains in agentic and coding benchmarks from re-post-training rather than architectural c…
Google DeepMind released Gemini Robotics 2, a set of three physical AI models for whole-body control, dexterous manipulation, and multi-robot collaboration, moving beyond table-top tasks to full humanoid control. The mod…
Prompt engineering, loop engineering, and graph engineering are three distinct units of control in AI systems, not competing techniques, according to a Marktechpost analysis. Prompt engineering controls one model respons…
Liquid AI released two open-weight bidirectional encoders, LFM2.5-Encoder-230M and LFM2.5-Encoder-350M, built on the LFM2 hybrid backbone with an 8,192-token context. The 350M model achieves a 17-task mean of 81.02, rank…
Microsoft AI released MAI-Cyber-1-Flash, a 137B-total-parameter cybersecurity model with 5B active parameters and 256k context length, designed for defensive tasks. Running inside Microsoft's MDASH multi-model agentic sc…
Moonshot AI's Kimi team and kvcache-ai have open-sourced AgentENV (AENV), a distributed platform for running agent environments at scale that powers agentic reinforcement learning (RL) training for Kimi K3, Moonshot's 2.…
Sakana AI released Fugu-Cyber, a cybersecurity-specialized orchestration model that reports 86.9% on CyberGym and 72.1% on CTI-REALM, surpassing reported scores for frontier models like GPT-5.5-Cyber and Claude Mythos Pr…
A developer trained a 6.4M-parameter decoder-only transformer from scratch in PyTorch on a single Colab T4 to power their recipe app Rasaveda, replacing external APIs with a custom model that runs inference on CPU. The m…
Datalab released Marker 2, a full rewrite of its open source document conversion pipeline, which scores 76.0% on the olmOCR-bench benchmark from Allen AI and sustains 2.9 pages per second on a single B200 GPU, outperform…
Datalab released Marker 2, a full rewrite of its open source document conversion pipeline, scoring 76.0% on the third-party olmOCR-bench benchmark from Allen AI while sustaining 2.9 pages per second on a single B200 GPU,…
Anthropic released Claude Opus 5, replacing Opus 4.8 as the Opus-tier flagship with unchanged pricing at $5 per million input tokens and $25 per million output tokens. The model scored 43.3% on FrontierBench v0.1 at max …
Andrew Ng released OpenWorker, an open-source desktop AI agent that produces finished deliverables instead of chat, running entirely on the user's machine. The four-layer architecture includes a Tauri 2 desktop shell, a …
Stanford PhD student Marcel Rød released Gigatoken, a Rust BPE tokenizer that encodes text at 24.53 GB/s on a 144-core AMD EPYC 9565 dual-socket setup, outperforming HuggingFace tokenizers by 989x and OpenAI's tiktoken b…