Matt Turck recommended LM Studio for local hosting.
I KEEP BLOWING THROUGH MY PERPLEXITY COMPUTER AND CLAUDE COWORK TOKENS
I HAVE SOME RESEARCH JOBS THAT I WANT TO RUN CONSTANTLY / HOURLY INDEFINITELY
NEED TO RUN LOCAL OPEN-SOURCE MODELS CONTINUOUSLY IN MY OWN PRIVATE CLOUD AT THIS POINT
TELL ME WHAT I SHOULD DO... @NousResearch TIME?
Many users praise Nous Research's fast shipping of reliable local models like Hermes for continuous research tasks, while some replies mock the poster's all-caps style.
No Digg Deeper questions have been answered for this story yet.
@Jason We have a surprise coming that will help with this
Stop. Its Nous Research time
I KEEP BLOWING THROUGH MY PERPLEXITY COMPUTER AND CLAUDE COWORK TOKENS
I HAVE SOME RESEARCH JOBS THAT I WANT TO RUN CONSTANTLY / HOURLY INDEFINITELY
NEED TO RUN LOCAL OPEN-SOURCE MODELS CONTINUOUSLY IN MY OWN PRIVATE CLOUD AT THIS POINT
TELL ME WHAT I SHOULD DO... @NousResearch TIME?
@Jason If you switch to Claude code, you can use these setting to use @akashnet models — you’ll barely notice a difference
@Jason posting in all caps makes you sound like an old man in my head
@Jason Purchase the ~100k worth of hardware needed to run GLM5.2.
@Jason Nous Research time
I KEEP BLOWING THROUGH MY PERPLEXITY COMPUTER AND CLAUDE COWORK TOKENS
I HAVE SOME RESEARCH JOBS THAT I WANT TO RUN CONSTANTLY / HOURLY INDEFINITELY
NEED TO RUN LOCAL OPEN-SOURCE MODELS CONTINUOUSLY IN MY OWN PRIVATE CLOUD AT THIS POINT
TELL ME WHAT I SHOULD DO... @NousResearch TIME?
@bmgentile @perplexity_ai I find they're both exceptional... i give Perplexity the edge on mobile
@zappstudios ok!
@Jason Should rly buy a few computers and add them to your http://actual.inc account
@Jason I just asked fable how to fix my prompt and tool usage for those things and he did a bang-up job fixing my prompts and tool usage for 10x reduction. It’s sort of cool watching a model struggle with how best to prompt a model.
@Jason use http://surplusintelligence.ai - inference marketplace with median discount of 40-60% off of Openrouter prices.
@Jason Llama.cpp with qwen3.6 27B using Hermes Agent as the harness. Will need 32G RAM so Mac Studio or really nice laptop/GPU. @UnslothAI quant is great. DMs open if you want to hop on a quick phone call/zoom to have me show you my setup.
@Jason Jason you were the local AI guy?… and you’re crying about cloud credits?
Self telling
@Jason Hello @lmstudio
I KEEP BLOWING THROUGH MY PERPLEXITY COMPUTER AND CLAUDE COWORK TOKENS
I HAVE SOME RESEARCH JOBS THAT I WANT TO RUN CONSTANTLY / HOURLY INDEFINITELY
NEED TO RUN LOCAL OPEN-SOURCE MODELS CONTINUOUSLY IN MY OWN PRIVATE CLOUD AT THIS POINT
TELL ME WHAT I SHOULD DO... @NousResearch TIME?
@Jason lmao nous research is evil. just go local and have a script that automates the local model for a prompt. you can do it via timed script or loop it with an llm.
@Jason https://www.microcenter.com/product/711961/amd-ryzen-ai-halo-developer-platform
anything beyond basic search is done better with computer. It’s having multiple agents/models handling the overall task and breaking it down. Where perplexity shines is that it has the ability to take the best use cases of different models and blend them into a single product. For example if a task needs to generate an image and also code up something it may use Claude for the code while tapping GPT or Gemini for the image. if you go to chat gpt or claude they would never use a competitors model even if the specific use case is superior elsewhere. Having an unbiased neutral platform in perplexity allows the orchestration of all models for all use cases.
@NousResearch @Jason @Jason 1000 percent setup Hermes Agent and trust Nous - one of the best decisions of my life and I've been on the bleeding edge using every harness, model, coding agent all day everyday. You will be happy you did.
@Jason Give your locals web tools and let em rip
@LvenderTempest @Jason yeah its a fairly universal human experience?