
Cohere's North Mini Code Now Runs Locally on a Mac at 67.6% SWE-Bench
Cohere's 30B MoE coding model now runs locally on a Mac via a 4-bit quantized build, with free API access on OpenRouter and native Ollama support
20 stories

Cohere's 30B MoE coding model now runs locally on a Mac via a 4-bit quantized build, with free API access on OpenRouter and native Ollama support

Anthropic's Python SDK now supports keyless authentication via Workload Identity Federation, letting CI and production workloads drop static API keys entirely

LiteParse v2.1 adds LLM-free markdown output that beats every open-source competitor across three standard benchmarks at 3ms per page

MCP's Enterprise-Managed Authorization goes stable, letting IT admins provision all Claude connectors org-wide through Okta with zero per-user OAuth setup

Claude Opus 4.7 autonomously programmed a robodog 20x faster than the best human team, signaling a new era of physical agentic AI

Devin Review now automatically scans every PR for logic-level security flaws and ships merge-ready fixes, going beyond what pattern-matching scanners can catch.

Perplexity's Brain gives its Computer agent a self-improving memory that gets cheaper and smarter with every task you run

Databricks adds Grok to Agent Bricks at DAIS 2026, giving enterprise teams a new frontier model option alongside their Lakehouse data

OpenAI's o3 Deep Research cracked 18 previously unsolvable rare pediatric disease cases in a peer-reviewed NEJM AI study with Boston Children's Hospital

Liquid AI releases two 350M retrieval models with sub-2ms GPU latency and best-in-class multilingual search across 11 languages

Google's BudouX brings ML-powered line breaking to CJK languages, now baked into Chrome, Android 14, and the ICU standard library

mem9 gives coding agents a shared, persistent memory layer that survives session resets, machine switches, and multi-agent workflows — built on TiDB Cloud
Kimi Work's new Goal Mode lets a local desktop agent pursue multi-step objectives autonomously, running round after round until the task is done -- no re-prompting required.

Mercury Agent is an open-source, permission-first AI agent with multi-agent orchestration, a Second Brain memory system, and 40+ hardened tools that runs 24/7 from your terminal or chat apps.

Tencent open-sources WeKnora, a Go-based platform that turns raw documents into a RAG pipeline, an agentic reasoner, and a self-maintaining Wiki — all self-hostable.

Ant Group's Robbyant open-sources LingBot-Depth, a masked depth modeling system that cuts indoor depth error by 70% for robotics and 3D vision

PaddlePaddle releases FastDeploy v2.5, adding Qwen3-VL support, W4AFP8 quantization, and 170+ bug fixes for production LLM serving

NVIDIA Warp 1.14.0 ships multi-environment FEM, CPU/GPU FFT fallbacks, pluggable logging, and a wave of autodiff fixes

Claudian brings full Claude Code agentic capabilities into Obsidian, turning your vault into an AI-powered working directory with file access, bash, MCP, and multi-agent support

Browser Harness gives LLMs raw Chrome DevTools Protocol access and lets agents write their own missing tools mid-task, replacing brittle automation frameworks with a 600-line self-healing harness.