Daily Digest – 2026-07-26

nestjs/nest

TypeScript framework with built-in DI, validation, and structure—less boilerplate than Express for scaling. (TypeScript, 76.2k★)

★ 76,214

juanfont/headscale

Self-hosted Tailscale control server—run your own mesh VPN without cloud vendor lock-in. (Go, 42.1k★)

★ 42,123

nushell/nushell

Shell with structured data pipelines instead of text—faster scripting for data-heavy automation. (Rust, 40.1k★)

★ 40,089

fastify/fastify

Low-overhead Node.js server framework—half the latency of Express when you need it. (JavaScript, 36.8k★)

★ 36,812

tailscale/tailscale

Mesh VPN built on WireGuard—easiest secure tunnel setup for remote teams, no router config needed. (Go, 34.4k★)

★ 34,353

venislathiya814092/faceswap_29

venislathiya814092/faceswap_29 (111 runs)

irulkenzei/voice-over-web

Coqui XTTS voice cloning: generate speech in any voice from text, no training required. (100 runs)

laguna-s-2.1

118B model for long-context reasoning—run locally if you have the VRAM and patience. (915 pulls)

medllama2

Llama 2 fine-tuned on medical data—cheaper than Claude for healthcare Q&A. (620.9K pulls)

opencoder

Code LLM in 1.5B and 8B sizes—local code completion without Copilot subscription. (610.3K pulls)

reflection

LLM trained to catch and correct its own mistakes—better reasoning without chain-of-thought hacks. (602.9K pulls)

nemotron

NVIDIA's Llama 3.1 70B—faster reasoning than stock Llama with tool use built-in. (598.8K pulls)

Salesforce/blip-image-captioning-base

Image-to-text model: turn screenshots, uploads into searchable captions without external API. (image-to-text, 1.8M downloads)

Qwen/Qwen3-ASR-1.7B

1.7B speech-to-text model—transcribe audio locally without Whisper API costs. (automatic-speech-recognition, 1.8M downloads)

jonatasgrosman/wav2vec2-large-xlsr-53-dutch

Dutch speech recognition model via wav2vec2—multilingual ASR without vendor APIs. (automatic-speech-recognition, 1.7M downloads)

handy-computer/nemotron-3.5-asr-streaming-0.6b-gguf

Tiny streaming speech model in GGUF—real-time transcription on edge devices. (automatic-speech-recognition, 1.7M downloads)

stable-diffusion-v1-5/stable-diffusion-v1-5

Image generation model—run locally instead of queuing on Midjourney or DALL-E. (text-to-image, 1.7M downloads)

princeton-nlp/SWE-bench_Verified

Curated dataset of real GitHub issues and fixes—benchmark your code LLM against actual bugs. (468K downloads)

WINGNUS/ACL-OCL

10K+ token classification examples—test NER and sequence labeling without manual annotation. (462K downloads)

cais/mmlu

14K multiple-choice questions across domains—standard benchmark for LLM evaluation. (459K downloads)

Brolly, a plain-text weather forecast site

Show HN: Brolly, a plain-text weather forecast site (215 upvotes) [hackernews_showhn]

I made some transistor animations

Show HN: I made some transistor animations (201 upvotes) [hackernews_showhn]

TouchGrass

Context-aware break reminders without invasive permissions (194 upvotes) [producthunt]

I mapped every US golf course

Show HN: I mapped every US golf course (176 upvotes) [hackernews_showhn]

Athena by Shoplazza

An orchestrator agent for your entire commerce stack (165 upvotes) [producthunt]

Speechius

The teleprompter that actually listens (127 upvotes) [producthunt]

Openbase

Manage your team of AI agents by voice, from anywhere (116 upvotes) [producthunt]

PureBox.ai

Review-first AI cleanup for your real Gmail inbox. (109 upvotes) [producthunt]