Daily Digest – 2026-08-13

apache/airflow

Schedule and monitor data pipelines without losing your mind to cron jobs and shell scripts. (Python, 46.5k★)

★ 46,465

calcom/cal.diy

Self-hosted scheduling that doesn't phone home or upsell you every Tuesday. (TypeScript, 47.5k★)

★ 47,519

rustdesk/rustdesk

Remote desktop that doesn't require trusting a third party with your screen. (Rust, 120.3k★ ⚡82★/day)

★ 120,322

lollipopkit/flutter_server_box

Monitor your servers from phone without opening twelve different dashboards. (Dart, 8.5k★ ⚡51★/day)

★ 8,471

psf/black

Stop arguing about Python formatting and just run it. (Python, 41.8k★)

★ 41,800

nautechsystems/nautilus_trader

Low-latency trading engine built in Rust for people who actually know what they're doing. (Rust, 25.5k★)

★ 25,460

prunaai/p-image-ideogram

Text-to-image at $0.003 per generation; cheaper than Midjourney if you're not picky. (176K runs)

littletry/compress-image

Batch image compression and conversion without opening Photoshop or a SaaS. (7K runs)

littletry/image-watermark

Add watermarks to images at scale; useful for protecting your content. (2K runs)

lightricks/ltx-2.5-fast

Generate video from text or images in seconds; text-to-video is finally practical. (676 runs)

irulkenzei/stable-audio-open

Generate audio from text prompts up to 47 seconds; BGM and sound design on demand. (231 runs)

falcon2

11B model trained on 5T tokens; solid baseline if you want something lightweight. (532K pulls)

notus

7B chat model fine-tuned on quality data; reasonable alternative to Llama for chat. (530.7K pulls)

notux

8x7B mixture-of-experts model; more capable than single models at similar token cost. (529.5K pulls)

firefunction-v2

Function calling model based on Llama 3; replace GPT-4 for tool use if you're hosted. (515.1K pulls)

deepseek-ocr

Vision model optimized for OCR; extract text from images without cloud APIs. (511.5K pulls)

agent-base

Base abstraction for HTTP agents; boring but essential for proxy support in Node. (249.7M/wk)

https-proxy-agent

HTTPS proxy support for Node without wrestling with Node's native limitations. (236.5M/wk)

http-proxy-agent

HTTP proxy support for Node; necessary if your requests need to route through a proxy. (133.3M/wk)

socks-proxy-agent

SOCKS proxy support for Node; useful for VPN routing or restricted networks. (55.4M/wk)

proxy-agent

Unified proxy agent that auto-detects protocol; saves you from hardcoding proxy logic. (29.0M/wk)

groq

Official Groq API client; fast inference if you need speed over feature parity. (6.5M recent downloads)

facebook/MusicGen

Generate background music from text prompts; great for not paying composers. (gradio, 5081 likes)

lmarena-ai/arena-leaderboard

Compare LLM performance across models without relying on vendor benchmarks. (static, 4969 likes)

microsoft/TRELLIS

Generate 3D models from text or images; useful if 3D modeling costs are killing you. (gradio, 4775 likes)

KingNish/OpenGPT-4o

Open-source multimodal model playground; beats paying OpenAI for experimentation. (gradio, 4306 likes)

nanotron/ultrascale-playbook

Practical guide for scaling LLM training without reinventing distributed computing. (static, 3977 likes)

nvidia/Qwen3.6-35B-A3B-NVFP4

Optimized Qwen model in FP4 quantization; run decent inference without renting GPUs. (text-generation, 11.3M downloads)

BAAI/bge-base-en-v1.5

Solid embedding model for semantic search without paying vendors per token. (feature-extraction, 10.9M downloads)

meta-llama/Llama-3.2-1B-Instruct

1B model you can actually run on consumer hardware; useful for offline features. (text-generation, 9.3M downloads)

deepseek-ai/DeepSeek-R1

Open reasoning model with actual chain-of-thought; worth testing if you're tired of OpenAI. (text-generation, 8.3M downloads)

Qwen/Qwen3-Embedding-0.6B

Tiny embedding model that runs locally and doesn't suck. (feature-extraction, 8.1M downloads)

KakologArchives/KakologArchives

Japanese text classification dataset; useful if your training data was thin. (2.4M downloads)

ryanmarten/OpenThoughts-1k-sample

Sample of reasoning traces for fine-tuning; helps if you're building reasoning models. (1.6M downloads)

ayuo/hd_tmp

High-definition dataset; details matter for certain computer vision tasks. (1.5M downloads)

Kane CLI

Natural language browser & mobile app tests from terminal (274 upvotes) [producthunt]

Ito

AI code review that runs your code (241 upvotes) [producthunt]

Nuphos

The AI-Native DevOps Workspace. (217 upvotes) [producthunt]

Human Behavior

Product analytics told you what happened. We handle it. (169 upvotes) [producthunt]

Scrimba Explain

Ask any question, get a video back instantly (153 upvotes) [producthunt]

Oasis

Where humans and agents come to work. (146 upvotes) [producthunt]

Skilldocs

Figma for markdown (139 upvotes) [producthunt]

Caveman

why use many token when few do trick (116 upvotes) [producthunt]