Inkling: Our Open-Weights Model

564 vimarsh6739 2026-07-15 18:46:32 UTC thinkingmachines.ai Qwen Share 137

The company released Inkling, a 975B-parameter Mixture-of-Experts model with 41B active parameters, supporting text, images, audio, and video, alongside Inkling-Small (12B active) for efficient customization via Tinker. Trained on 45T tokens, Inkling offers controllable reasoning, safety features, and is available on platforms like Hugging Face and Tinker, with partnerships for deployment and ...

SQLite should have (Rust-style) editions

45 gnyeki 2026-07-15 22:46:32 UTC mort.coffee Qwen Share 22

SQLite is praised for its embedded use and self-contained design but faces challenges with ROWID reuse, flexible typing leading to data validation issues, and concurrency limitations. Solutions like strict tables, WAL mode, and an edition system for updated defaults are proposed to address these issues while maintaining compatibility.

Grok Build is open source

190 skp1995 2026-07-15 20:46:32 UTC github.com/xai-org Qwen Share 218

Grok Build is SpaceXAI's terminal-based AI coding agent with a TUI for codebase interaction, file editing, shell commands, and task management, supporting macOS, Linux, and Windows. It uses Rust, offers CLI installation, and is Apache 2.0 licensed with third-party notices.

The Anti-Mac User Interface (1996)

22 ninglor 2026-07-15 22:54:32 UTC nngroup.com Qwen Share 2

The article explores alternative human-computer interfaces by violating Macintosh design principles, arguing that metaphors like the desktop limit innovation. It advocates for interfaces leveraging language, richer object representations, and shared control to transcend conventional WIMP models.

Speculative Growth and the AI "Bubble" [pdf]

46 johnbarron 2026-07-15 21:46:32 UTC economics.mit.edu Qwen Share 39

High AI valuations may reflect a third possibility: temporary overvaluation drives capital accumulation, shifting income to savers and lowering long-term interest rates, even if valuations later correct. This creates a self-sustaining high-capital economy with higher wages and lower interest rates, but risks failure if corrections occur too early.

Governments, companies, nonprofits should invest in free, open source AI [pdf]

47 bilsbie 2026-07-15 21:46:32 UTC siegelendowment.org Qwen Share 13

The article reflects on the 1980s shift from proprietary to open-source software, crediting Richard Stallman's advocacy for free knowledge-sharing as foundational to tech progress. It warns that AI's increasing closure risks stifling innovation and transparency, urging open access to models and data to preserve scientific and societal advancement.

Stripe and Advent have made a joint offer to acquire PayPal – sources

312 rvz 2026-07-15 15:46:32 UTC reuters.com Full Share 185

Please enable JS and disable any ad blocker

Metal-Organic Frameworks, Chemistry's New Miracle Materials

12 andsoitis 2026-07-15 23:02:32 UTC chemistry.berkeley.edu Qwen Share 2

Metal–organic frameworks (MOFs), ultra-porous materials with vast surface areas, excel at trapping gases, harvesting water from arid air, and delivering drugs. Pioneered by Omar Yaghi, they offer scalable, solar-powered solutions for water scarcity, carbon capture, and medicine, with over 20,000 variants now enabling diverse environmental and medical applications.

LLM Networking with MikroTik

12 gregsadetsky 2026-07-15 22:46:32 UTC blog.greg.technology Qwen Share 3

The author uses LLMs to streamline MikroTik network setups, leveraging their efficiency while cautioning against errors, and shares tips like REST API use, CAPsMAN, and MAC-Telnet for smoother, version-controlled configurations. Networking's complexity is acknowledged, balancing LLMs' chaotic help with meticulous testing and documentation.

Show HN: Firefox in WebAssembly

99 coolelectronics 2026-07-15 21:46:32 UTC developer.puter.com Full Share 49

The Gecko engine compiled to WebAssembly. The real Firefox interface, rendering entirely inside this browser tab. GPU acceleration WebGL-based rendering for web content. Enable JIT Experimental JS → WASM JIT. Expect bugs. Wisp server WebSocket proxy that tabs load web content through. Networking in the browser powered by Puter.js

Show HN: One More Letter

9 hmate9 2026-07-15 23:12:32 UTC playonemoreletter.com Full Share 3

One More Letter is a daily word game where every round adds one letter and rearranges them into a new answer.

Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

216 neomindryan 2026-07-15 15:46:32 UTC neomindlabs.com Qwen Share 139

A 13-year-old HP server with Ivy Bridge Xeons runs Google's 26B-parameter Gemma 4 model at ~5 tokens/second via CPU-optimized code, overcoming AVX2 instruction set limitations with custom patches. The project highlights adapting modern AI to legacy hardware through engineering, not just subscriptions, enabling local inference on outdated systems.

Duskers, the scary command line game, is getting a sequel

83 spacemarine1 2026-07-15 19:46:32 UTC elbowgreasegames.substack.com Qwen Share 12

Misfits Attic announced *Duskers 2.0*, funded by Max McGuire’s Stray Signal, during PC Gamer’s event, expanding the survival game’s narrative with moral choices and deeper gameplay. EGG’s community-driven funding model, supporting indie developers like Tim through prototypes and partnerships, highlights a new era of collaborative game development.

Book prizes don't work how you think

44 samclemens 2026-07-15 20:46:32 UTC rebeccamakkai.substack.com Qwen Share 21

Rebecca Makkai reveals book prize judging relies on small, subjective panels with no hidden agendas, emphasizing randomness, personal taste, and the emotional weight of decisions. The process, shaped by limited reading time and chance, impacts authors' careers while defying myths about bias or secret criteria.

P2P local file transfer based on WebRTC

12 halb 2026-07-15 22:46:32 UTC pairdrop.net Full Share 9

Instantly share images, videos, PDFs, and links with people nearby. Peer2Peer and Open Source. No Setup, No Signup.

Brainless: Shadcn components that look like Claude Code, Codex and Grok

73 benswerd 2026-07-15 19:46:32 UTC brainless.swerdlow.dev Full Share 13

Claude Code, Codex, and Grok interfaces as shadcn components.

Command Line Interface Guidelines

42 subset 2026-07-15 20:46:32 UTC clig.dev Qwen Share 2

This open-source guide modernizes UNIX principles for CLI design, focusing on human-centric usability, composability, and clear documentation. It provides actionable best practices for intuitive interfaces, error handling, and output formatting to improve accessibility and efficiency.

Collection of Digital Clock Designs

162 levmiseri 2026-07-15 16:46:32 UTC clocks.dev Full Share 34

A collection of thoughtfully designed digital clocks.

Artie (YC S23) Is Hiring Software Engineers

0 2026-07-15 17:46:32 UTC jobs.ashbyhq.com Full Share

Artie Jobs

Nul Characters in Strings in SQLite

5 basilikum 2026-07-15 23:05:32 UTC sqlite.org Prefix Share 0

SQLite allows NUL characters (ASCII 0x00, Unicode \u0000) in the middle of string values stored in the database. However, the use of NUL within strings can lead to surprising behaviors: The .dump command in the CLI omits the first NUL character and all subsequent text in the SQL output that it generates. In fact, the CLI omits everything past the first NUL character in all contexts. CREATE ...

Mysteries of Telegram Data Centers (2022)

236 theanonymousone 2026-07-15 13:46:32 UTC dev.moe Full Share 122

Voxatron

46 lsferreira42 2026-07-15 19:46:32 UTC lexaloffle.com Full Share 15

Enter your email address below, choose a payment method, and you will be emailed a download link. Purchasing Voxatron also gives you access to all future updates. If you have any trouble, please feel free to email us.

Show HN: E– – A language you dial between English and Python

5 OdedF 2026-07-15 22:46:32 UTC github.com/frmoded Prefix Share 3

A programming language you write in plain, canonical English — and that compiles deterministically to Python. E-- ("English--") is English with the ambiguity removed: a closed grammar and a fixed vocabulary, with exactly one canonical phrasing per construct. It is meant to read and edit like English while still compiling to ordinary, reproducible Python. LLM-generated code is fuzzy at ...

Prioritize mental health, and why communication is so important

271 ramon156 2026-07-15 11:46:32 UTC ramones.dev Qwen Share 235

Struggling with severe depression and career setbacks, the user acknowledges personal challenges in motivation and discipline, now focusing on healing with support. They seek stability, improved work discipline, and meaningful contributions while navigating mental health.

Designing APIs for Agents

35 benswerd 2026-07-15 19:46:32 UTC freestyle.sh Qwen Share 18

Designing APIs for agents prioritizes clarity, explicitness, and precise error handling over human-centric utilities, as agents require unambiguous documentation and direct code interpretation to function effectively without relying on defaults or abstractions. The author advocates for minimalistic, self-explanatory APIs with explicit field names and comprehensive guides, emphasizing that ...

We don't use AI in any of our design or production processes

83 tony_cannistra 2026-07-15 21:46:32 UTC mass-driver.com Qwen Share 71

Typography evolved through centuries of human craftsmanship, embedding cultural and physical experiences. AI, lacking human nuance and cultural depth, risks stagnating innovation and marginalizing underrepresented voices in design.

MITS: Rockets, Calculators, and Personal Computers

24 BirAdam 2026-07-15 19:46:32 UTC abortretry.fail Qwen Share 1

Henry Edward Roberts founded MITS and created the Altair 8800, the first successful personal computer, which sparked the PC revolution and inspired Microsoft and the S-100 bus standard. His vision and innovations laid the groundwork for modern computing, though he later pursued farming and medicine as originally intended.

Show HN: misa77 - a codec that decodes 2x faster than LZ4 (at better ratios)

122 nonadhocproblem 2026-07-15 16:46:32 UTC github.com/welcome-to-the-sunny-side Qwen Share 39

misa77 is an LZ-based codec optimized for write-once, read-many scenarios, offering constant memory use (<5 MB) and faster decompression for highly compressed files, with experimental modes enhancing CPU efficiency. It balances compression ratios and throughput, excelling in decode speed for compressible data while supporting CLI-driven compression and parameter tuning.

Towards a harness that can do anything

161 evakhoury 2026-07-15 14:46:32 UTC eardatasci.github.io Qwen Share 81

The author advocates for transparent, Unix-inspired LLM harnesses that minimize cognitive load by leveraging the model's existing systems knowledge, modular tools, and event-driven architectures like Ambiance to enable efficient, self-directed agent workflows. Key principles include deterministic goal pursuit, flat text interfaces, structured data storage, and robust logging to ensure ...

Show HN: Low-latency local LLM runner via OpenJDK Panama FFM (Java 22)

21 KingJoker 2026-07-15 19:46:32 UTC github.com/projectargus-cc Qwen Share 3

libargus is a zero-allocation native AI runtime consolidating Vision, Speech, and LLM pipelines via Project Panama FFM, built on GGML/llama.cpp for high-performance, thread-safe execution. It enables JVM integration with Java bindings, supports multimodal processing, and offers zero-copy memory, KV cache quantization, and CUDA acceleration for low-latency inference.