新秀榜（30天）

100% local video dubbing on your desktop — dub videos in your own voice. Private AI dubbing with no cloud, no uploads, no time-stretching.

Streaming ASR, audio captioning and speech QA over a whisper-style encoder with pluggable LLM backends

Production-grade recipe for DeepSeek-V4-Flash-0731 (284B MoE) on 2x NVIDIA DGX Spark: self-healing 2-node vLLM cluster, reboot-verified, tuned DSpark speculative decoding (~75 tok/s), full benchmarks, OpenAI Codex CLI integration

OpenAI-compatible reverse proxy for free multimodal AI APIs (chat / images / videos / audio / 3d)

Writing that's kin to yours

Local-first, capability-aware, traceable audio and video transcription Skill for Codex

Proxima lets existing GPUs serve 4x more concurrent requests

Network-faithful simulation of LLM serving and training deployments

997 playable, source-attributed MiniMax H3 / Hailuo H3 video cases, 309 complete public Prompts, and 13 source-checked guides.

MLOps — vLLM + LiteLLM + Portainer CE + Netdata. Case study song ngữ: kiến trúc, metrics, bài học vận hành. · Self-hosted LLM inference done the MLOps way, a bilingual case study

本地优先的 macOS 与 Windows 语音输入工具，支持端侧 SenseVoice 识别、全局语音输入、文件转写及可选 AI 校对。Local-first voice dictation for macOS and Windows with on-device SenseVoice, typing into any app, file transcription, and optional AI proofreading.

Local LLM speed measurements from a 4x RTX 3090 rig, with exact launch commands and per-run provenance.

A low-token visual evidence compiler for text-only coding agents. Convert images into compact Visual Evidence Packets (VEP) for DeepSeek, Codex, Claude Code, and other text-only models.

Codex Skill：让纯文本模型（DeepSeek）借助 StepFun step-3.7-flash 获得看图能力 | Give text-only Codex models (DeepSeek) image understanding via StepFun step-3.7-flash

Thinking-native compiler for LEET Language Understanding item sets — killer-item design grammar, verification separation, deterministic gate chain, 3-model benchmark

Hikari's curated English measured notes on local LLM systems (serving, specdec, training, steering).

Introducing Muse Glimmer: open-weight 30B agentic multimodal model that runs on your device (Meta). Interactive local agent lab + guide. Apache 2.0 · on-device AI · function calling

Run DeepSeek-V4-Flash 304B on a single NVIDIA GB10 / DGX Spark - 2-bit MoE planes, FP4 quality recovery, speculative decoding

Move files between devices with a screen and a camera — offline optical transfer of RVF cognitive containers and WASM artifacts over animated QR codes. Mobile-first, no network, no accounts, SHA-256 verified.

MCP server that gives text-only LLM coding agents vision — analyze images via any multimodal model (mimo, Claude, Gemini, OpenAI-compatible). Works with Claude Code, Codex, Kimi, opencode, PI.