Daily Digest – 2026-07-19

ClickHouse/ClickHouse

Real-time analytics database that handles billions of rows without breaking a sweat. (C++, 48.7k★)

★ 49,026

swiftlang/swift

Apple's language for iOS/macOS that doesn't require years of Objective-C therapy. (Swift, 70.2k★)

★ 70,166

KRTirtho/spotube

Spotify client that works without paying Spotify, streams from YouTube instead. (Dart, 47.7k★)

★ 48,401

vinta/awesome-python

Curated Python tools list—bookmark it, stop asking "what package should I use." (Python, 309.1k★ ⚡203★/day)

★ 309,115

ripienaar/free-for-dev

Exhaustive list of free tiers so you can build without hitting your credit card. (HTML, 129.8k★ ⚡107★/day)

★ 131,373

ccxt/ccxt

Unified API for 100+ crypto exchanges—trade everywhere without learning 100 APIs. (Python, 43.3k★)

★ 43,480

paperless-ngx/paperless-ngx

Self-hosted document scanner that actually indexes PDFs so you can find stuff. (Python, 43.2k★ ⚡40★/day)

★ 43,936

bytedance/seedance-2.0

Video generation with audio sync and control—multimodal without the OpenAI price tag. (1.2M runs)

rafaelgalle/whisper-diarization-advanced

Transcription + speaker diarization for messy audio—call centers finally get relief. (598K runs)

google/gemini-3.1-flash-tts

Google's TTS with 30 voices and 70+ languages—faster than ElevenLabs for most uses. (315K runs)

leonardoai/lucid-origin

Image generation with better prompt adherence—stop fighting the model for quality. (281K runs)

google/gemini-3.5-flash

Fast multimodal reasoning from Google—cheaper than GPT-4o for most tasks. (193K runs)

internlm2

InternLM2.5 7B locally—reasoning model for edge devices without API dependency. (968.1K pulls)

samantha-mistral

Mistral-based conversational model—personality without the token overheads. (966.7K pulls)

llama3-groq-tool-use

Llama3 with native function calling—agents that reliably invoke tools. (962.2K pulls)

granite3.2-vision

2B vision model for document understanding—extract tables and charts without paying APIs. (946.2K pulls)

xwinlm

Llama 2-based model competitive on benchmarks—solid general-purpose local alternative. (944.4K pulls)

semantic-release

Automated semver versioning and publishing—deploy without manually tweaking package.json. (2.6M/wk)

@appium/types

Various type declarations used across Appium (2.4M/wk)

@appium/schema

Appium Configuration Schema (2.2M/wk)

@appium/support

Support libs used across Appium packages (2.2M/wk)

selenium-webdriver

Official Selenium WebDriver for JavaScript—the standard for browser automation. (1.8M/wk)

pydantic

Type validation with Python hints—structure your data before it breaks production. (246.9M recent downloads)

litellm

Unified LLM API—switch between OpenAI, Claude, Bedrock without rewriting code. (170.5M recent downloads)

tokenizers

Fast tokenization for transformers—batch encode without bottlenecking inference. (49.7M recent downloads)

transformers

HuggingFace's library for every ML model—the de facto standard for NLP and vision. (42.9M recent downloads)

playwright

Cross-browser automation faster than Selenium—test Chromium, Firefox, Safari with one API. (21.1M recent downloads)

r3gm/wan2-2-fp8da-aoti-preview

Text-to-speech model playground—test voice outputs without building locally. (gradio, 2907 likes)

not-lain/background-removal

Background removal in your browser—no Photoshop, no paid APIs, just works. (gradio, 2894 likes)

mrfakename/E2-F5-TTS

Fast text-to-speech model that sounds natural without paying ElevenLabs monthly. (gradio, 2888 likes)

sanchit-gandhi/whisper-jax

Whisper speech-to-text on JAX—faster inference if you're already in the JAX ecosystem. (docker, 2836 likes)

HumanAIGC/OutfitAnyone

Virtual try-on for clothing—generate outfit previews without hiring a photo shoot. (gradio, 2812 likes)

Qwen/Qwen3-32B

Open 32B model that rivals GPT-4 on reasoning, runs locally or via API. (text-generation, 7.6M downloads)

Qwen/Qwen3-1.7B

Tiny 1.7B model for edge devices—fast inference when you can't afford latency. (text-generation, 5.8M downloads)

openai/whisper-large-v3

Latest Whisper for speech-to-text—handles accents and noise better than v2. (automatic-speech-recognition, 5.8M downloads)

farbodtavakkoli/OTel-LLM-E4B-IT

Fine-tuned model for telecom RAG—production-ready if your domain is telecom. (text-generation, 5.5M downloads)

hmellor/tiny-random-LlamaForCausalLM

Tiny random Llama for testing pipelines without downloading gigabytes. (text-generation, 5.5M downloads)

openai/gsm8k

Math reasoning benchmark dataset—evaluate if your model can actually do arithmetic. (926K downloads)

k9cli/video-vec2wav2-tokenizer

video-vec2wav2-tokenizer (910K downloads)

HennyPr/ps2_hf2

ps2_hf2 (739K downloads)

ZooData

The data layer for AI agents (619 upvotes) [producthunt]

Clark

An AI coworker with its own cloud computer (490 upvotes) [producthunt]

I replaced a $120k bowling center system with $1,600 in ESP32s

Show HN: I replaced a $120k bowling center system with $1,600 in ESP32s (477 upvotes) [hackernews_showhn]

OpenSEO

The open source Ahrefs alternative (471 upvotes) [producthunt]

LiveDemo

Open-source alternative to Storylane, Navattic, and Arcade (345 upvotes) [producthunt]

Mirage

Turn your SaaS into a clickable demo in 90 seconds. (246 upvotes) [producthunt]

Spycost

Fell for a discount again? (244 upvotes) [producthunt]

OpenMarkdown

A markdown editor you and your agent co-edit (205 upvotes) [producthunt]