Fetching from the wire…
01
02
03
04
05
06
07
08
Models2026-09-05 · source-backed
LLMPSP runs Falcon-H1-Tiny-90M-Instruct on PSP-2000/3000/Street/GO hardware with custom firmware, using a bespoke FHQ4 4-bit format tuned for memory-stick access and a 52MB allocation, reaching 0.5 to 0.6 tok/s, so a reply takes one to three minutes. The model is a hybrid of grouped-query attention (8 query heads, 2 KV heads) and Mamba2 recurrent branches across 24 layers, accelerated on the PSP's VFPU. Code is MIT; the PSP-1000's 32MB of RAM isn't enough (GitHub).
Each link below shares sources, entities, or timing with this story.
No compiled code. No runtime. No dependencies. Just 18 markdown files in a .claude directory. Matt Pocock's skills repo hit 75,700 stars this week, gaining 6,400 in seven days. It's the #1 trending AI repo on GitHub. The repo solves four problems that every Claude Code user hi...
QM went up under MIT license. Created July 29. As of the GitHub API check: 8,420 stars, 887 forks. Five days. YC uses it internally across accounting, legal, events, and engineering, including to build QM itself. Every employee and every Slack room gets its own scoped memory,...
affaan-m/ECC (36.3k forks, MIT) bundles 67 agents, 284 skills, 94 legacy command shims, and "instincts", patterns learned from prior sessions with confidence scores that auto-recall when relevant, plus a .ecc/memory/ markdown vault that's explicitly cross-harness, so context s...
Blaizzy/nativ (1,163 stars, Swift, MIT, macOS 26+) comes from the mlx-vlm author and bundles that server into a SwiftUI app that discovers MLX models already in your HF cache. It exposes OpenAI-compatible chat, Responses, image, audio and model endpoints plus Anthropic Message...
AMAP-ML/LongHorizon-Harness (370 stars, MIT, paper at arXiv 2608.01964) splits long computer-use work into three roles with independently assignable models. Reported: WeaveBench 51.8% → 80.7%, Terminal-Bench 2.1 69.7% → 77.2%, OSWorld 2.0 2.8% → 8.3%. It integrates with Claude...
I check Product Hunt maybe once a week and usually regret it. Today's board is worth reading as market structure. The July 30 leaderboard: SKI at 277 upvotes (free voice input for Claude Code and Codex). AI Search Console at 249 (prompt analytics and citation mapping). Memmy A...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.