search

assafelovic / gpt-researcher

LLM based autonomous agent that does online comprehensive research on any given topic
https://gptr.dev
Apache License 2.0
14.25k stars 1.86k forks source link

Please add support for pinokio.computer #562

Open xenstar opened 4 months ago

xenstar commented 4 months ago

Can you create a config file for pinokio.computer, so we can easily host this on our computer, and use openrouter api aswell. It would be great then :)

assafelovic commented 3 months ago

Hey @xenstar thanks for the feedback! Can you explain more about the use case for this? And why pinokio vs other alternatives?

xenstar commented 3 months ago

Hi @assafelovic

Well, for https://pinokio.computer, they have premade scripts, so you can just submit your scripts. Users have to search for it and then click on install. It will automatically download everything needed to run the script.

As for openrouter.ai, there are a lot more models available at a cheaper price, so we can run GPT researcher at a cheaper price. We can even use Perplexity directly.

Details here - https://program.pinokio.computer/#/

jay377 commented 3 months ago

pinokio.computer

Damn this looks cool thanks for the ref.

EDIT sorry Xen i thought you where saying no to openrouter, bu actually recommending. Absolutely YES PLEASE to openrouter they have the cheapest models in the world all in 1 place and an easy single API, there is nowhere or anything better ive spent days making sure, I support this 100% please @assafelovic

xenstar commented 3 months ago

Openrouter is indeed the cheapest, as it has many providers available and routes to the cheapest one.

For openai, this one provides cheaper pricing than the official open api - https://deepbricks.ai/pricing

jay377 commented 3 months ago

Openrouter is indeed the cheapest, as it has many providers available and routes to the cheapest one.

For openai, this one provides cheaper pricing than the official open api - https://deepbricks.ai/pricing

I use openrouter but add the provider lock code to python to ground me to together ai only for llama 70b and that gives me the fastest (together way is faster then the rest yet they split the calls between if you dont lock it, so no thanks) cheapest llama 70b you can find without any rate limits basically, its awesome. grok it doesnt count because of that rate limit but its fine if you dont need vol

Holy hell now i goto eat my words cause deep bricks is even cheaper for llama 3 wow, but speed, thats im wondering..