Sat, Jul 18
HomeAboutSubscribe

SIGNAL

Saturday, July 18, 2026
18 stories · 5 min read

★ Must ReadInkling, Gemma 4 Updates, and 1-Bit Qwen3.6

Kaitchup — Inkling, Gemma 4 Updates, and 1-Bit Qwen3.

AI #177 Part 2: Wish You Were Here
Zvi Mowshowitz

As usual, part 2 of the weekly deals with speculative, regulatory, political and alignment questions.

Source →
vs
The state of open source AI
Hacker News

Trending on Hacker News with 382 points and 281 comments.

Source →

[AINews] Kimi K3 2.8T-A50B: the largest open model ever released; Opus 4.8-class at Sonnet 5 pricing

Latent Space — [AINews] Kimi K3 2. 8T-A50B: the largest open model ever rele...

🔬 The Lab of the Future Should Feel Like a Data Center — Andy Beam & Rafa Gómez-Bombarelli, Lila Sciences

Lila is betting that science, not the internet, is the last untapped source of training data. We went to find out what that actually looks like in a room full of robots.

Analysis: Stripe bids $53B for PayPal; and the DeFi Post-Mortem

Lex Fridman — Analysis: Stripe bids $53B for PayPal; and the DeFi Post-Mor...

★ Must ReadKimi K3, and what we can still learn from the pelican benchmark

Plus reverse-engineering Grok Build, Doom in SQLite, and how Codex generates pets

Qwen3.6-27B KV Cache Quantization in vLLM: Accuracy, Memory, and Speed

A smaller KV cache enables longer sequences and higher concurrency with virtually no loss in accuracy.

LAI #134: Your First LLM App on AWS for Under a Dollar

Loop engineering with Claude Code, plus a Towards AI enterprise launch, vLLM on L40S, and context windows as memory management

★ Must Read[AINews] Thinky's Inkling: 975B-A41B multimodal, new best American Apache 2.0 open model (with Inkling-Small, 276B-A12B)

Thinky's first full LLM release is a banger and bonus: it's open weights!

Inkling, Gemma 4 Updates, and 1-Bit Qwen3.6
Benjamin Marie, Kaitchup
Databricks hits $188B valuation, extending its run as AI’s favorite second act
Julie Bort, TechCrunch AI
TikTok is testing an AI likeness detection tool
Jay Peters, The Verge AI