Telexed

AI news through a solo-operator lens — only what changes your day5 of 843

FILTER[All][Agents & tools][Models & API][Generative media][Infra & SaaS][ASO & growth][Indie business][Idea signals][Other][★6+ high-signal]

GitHub Trending Weekly ✕clear filters

Fri, May 291 dispatches

#0843
#0843Generative media GitHub Trending Weekly2 days ago
`MoneyPrinterTurbo`: open-source one-click short video generator
60radar
MoneyPrinterTurboOpen-source video generator — turns keywords into short videos
A topic or keyword can produce script, footage, subtitles, music, and a finished HD video. Useful for fast promo-video experiments, but quality and asset sourcing still need review.
- Generates copy, media, subtitles, BGM, and final video from one prompt; Web UI and API both exist for workflow integration.
- Supports 9:16 1080x1920 and 16:9 1920x1080, so the same pipeline can cover Shorts/Reels/TikTok and YouTube-style clips.
- Batch generation creates multiple versions at once; practical for testing hooks, narration, and pacing before choosing one output.
- LLM providers include OpenAI, Gemini, DeepSeek, Ollama, Moonshot, Azure, Qwen, and more; provider choice affects cost, latency, and availability.
- Local setup supports uv sync --frozen, Docker, and Windows one-click package, but video processing gets smoother with more CPU/RAM/GPU.
Source: github.com/harry0703/MoneyPrinterTurboRead original →
FIG-8431:1
60radar
FIG-8431:1

Wed, May 272 dispatches

#0842
#0842Generative media GitHub Trending Weekly3 days ago
`SANA` expands into image, video, and controllable world-model generation
70radar
SANAGenerative media model stack — optimized for high-res image and video
A research repo has become a full training/inference stack for high-res media. Useful for custom pipelines, but heavy for quick SaaS integration.
- SANA-WM adds 720p, 1-minute video with 6-DoF camera control; strong for simulation and controllable scene generation ideas.
- SANA-Video supports text-to-video and image-to-video, with LTX-VAE and LTX2 Refiner paths up to 2K output.
- SGLang support exposes high-performance serving through an OpenAI-compatible API, making product integration less painful.
- ComfyUI, Hugging Face, diffusers, and training recipes are all mentioned; the repo is closer to a platform than a single model drop.
Source: github.com/NVlabs/SanaRead original →
FIG-8421:1
70radar
FIG-8421:1
#0841
#0841Generative media GitHub Trending Weekly4 days ago
`Presenton`: open-source AI presentation generator and API
80radar
PresentonAI presentation generator — self-hosted with editable PPTX export
Self-hosted deck generation with BYOK, local models, and editable PPTX export. Strong fit when slide output must be automated without SaaS lock-in.
- Runs via Docker or desktop apps for macOS, Windows, and Linux — useful for keeping documents and prompts off third-party SaaS.
- Model support spans OpenAI, Gemini, Claude, Bedrock, Ollama, LM Studio, and custom endpoints, so cost and privacy are configurable.
- Exports editable PPTX and PDF, not just static images. That matters for client decks, sales material, and reusable templates.
- Includes an AI presentation generation API and built-in MCP server, making slide creation scriptable from agents or internal tools.
- Apache 2.0 license plus custom HTML/Tailwind templates lowers lock-in; the tradeoff is hosting and template maintenance work.
Source: github.com/presenton/presentonRead original →
FIG-8411:1
80radar
FIG-8411:1

Fri, May 222 dispatches

#0840
#0840Generative media GitHub Trending Weeklylast week
`ViMax`: Agentic Pipeline for End-to-End Video Generation
60radar
ViMaxAI video tool — automates script-to-final-video workflow
It packages scriptwriting, storyboarding, character design, and video generation into one workflow. Useful as a product-direction signal, but production value depends on model quality.
- Idea2Video, Novel2Video, Script2Video, and AutoCameo cover four input modes, turning raw concepts, novels, scripts, or photos into video workflows.
- The core promise is multi-agent orchestration across director, writer, producer, and generator roles, not just another clip generator.
- It targets common AI video pain points: short outputs, character inconsistency, missing script/audio/story structure.
- Best near-term use is prototype content pipelines, landing-page demos, and creator-tool experiments; reliability still needs hands-on validation.
Source: github.com/HKUDS/ViMaxRead original →
FIG-8401:1
60radar
FIG-8401:1
#0839
#0839Generative media GitHub Trending Weeklylast week
`Supertonic`, Open-Weight On-Device TTS With ONNX Runtime
80radar
SupertonicOn-device TTS model — local 31-language synthesis via ONNX
Local speech generation now covers browser, mobile, desktop, and edge without an API call. The OpenAI-compatible local server makes drop-in narration features much easier.
- 99M parameters keeps downloads, cold starts, and memory use below larger 0.7B-2B open TTS systems.
- Supports 31 languages plus lang="na", so multilingual apps can skip separate language detection for many flows.
- Python SDK v1.3.1 adds supertonic serve with /v1/audio/speech, useful for replacing hosted TTS in local dev or privacy-sensitive apps.
- Runtime examples cover Node.js, Browser WebGPU, Swift, iOS, Flutter, Rust, Go, and C++; integration paths are unusually broad.
Source: github.com/supertone-inc/supertonicRead original →
FIG-8391:1
80radar
FIG-8391:1

`MoneyPrinterTurbo`: open-source one-click short video generator

`SANA` expands into image, video, and controllable world-model generation

`Presenton`: open-source AI presentation generator and API

`ViMax`: Agentic Pipeline for End-to-End Video Generation

`Supertonic`, Open-Weight On-Device TTS With ONNX Runtime