Agents A1 + Hermes,Zed: This LOCAL AI CODER is CRAZY!

In this video, I'll be telling you about Agents A1, a new 35B Mixture-of-Experts agentic model from InternScience that runs locally, supports long context, performs really well on tool use and instruction following, and could be one of the best local AI models right now. -- Key Takeaways: 🚀 Agents A1 is a 35B Mixture-of-Experts agentic model built by InternScience. ⚡ Even though it has 35B total parameters, only around 3B are active at a time, making it fast for local use. 📚 The model supports a massive 256K token context window for long documents and conversations. 🧠 Agents A1 performs strongly on benchmarks like HLE, Seal Zero, GAIA, BrowseComp, IFBench, and IFEval. 🔍 It is especially impressive at search, tool use, structured output, and instruction following. 💻 The model works well for local coding, file editing, summarization, research, and agentic workflows. 🛠️ You can run it through LM Studio, Ollama, llama.cpp, SGLang, or vLLM with OpenAI-compatible endpoints. 🔒 Since it runs locally and is Apache 2.0 licensed, it is a great option for privacy-friendly AI workflows. 👍 Overall, Agents A1 looks like one of the most exciting local agentic models available right now.