Put "Liangzi" into MacBook: Actual test of local AI voice chat, the delay is reduced to 1.3 seconds
In this issue, "Liangzi" is installed in a MacBook with an M5 chip and 16GB of memory: an AI voice assistant that can listen, think, and speak with Liangzi's voice. From the time I finished speaking to the time it spoke, it actually took about 1.3 seconds. The entire system is broken down into five building blocks: • VAD judgment: Silero + SmartTurn, two small models run on the CPU…
Top AI LLMs for Every Use Case
Try Gemini 3.1 Pro FREE 👉 → Writing & Research: GPT-5.4, Claude 4.6, Gemini 3.1 Pro, Perplexity → Social Content: Grok 4, GPT o3, DeepSeek → Academic / STEM: Claude Opus 4.6, MiniMax M2.7 One place. The right AI
In this video, I'll be telling you about Free Claude Code, also known as FCC
Which lets you use Claude Code for free by connecting it to different model providers like OpenRouter, Gemini, NVIDIA NIM, DeepSeek, Groq, Cerebras, Kimi, GLM, Ollama, and more through a local proxy.
Run E2E with Deepseek multi-modal + Midscene, the speed is almost human... (The video is not accelerated)
You have reached the end of the archive
All of deepseek