- It is not super fast for me either so far.
- It is probably because of ollama, or?
- What else have you been using?
Discussion
I think it’s just slow because of hardware. You need a beast computer for it to be fast. I’m just getting started since I only recently got a machine good enough to run models locally. Msty works great for using Ollama plus tools like web search. Been using gemma3 for that. I don’t think you can use MCP servers with Msty desktop though.
How would you compare msty with goose?
I just started playing with Goose today, but I think I’ll use Goose more for coding and automating tasks and Msty for general private chat with web search. Not sure if I’m even going to use Goose though. Still feeling it out. Also, I don’t use Msty much. I still mostly use hosted solutions such as duck.ai and perplexity.
I use those 2 as well. Perplexity is quite nice. Although really stupid sometimes.
I am thinking using goose for making automation from voice prompts. I have so much time the problem, that I have an idea what could be implemented, but I dont have the pc in hand.
And I want to have better control over my vibe code environment.
I’m trying out goose with Claude 3.7 sonnet now. It’s way better and faster of course. One shotted a decent breakout game in a single html file with no dependencies as a test. Total cost was 15 cents. I chose 3.7 because I thought it was cheaper than 4.0, but looks like it’s the same price.
Do you use it through ppq?
Is there a way to do that with goose? I just made an Anthropic API key and loaded my account with $5 to test it out.
I think so. Through open ai api, you shall be able to configure.
That is my plan to use claude
Have you played around with goose since? What is your thought on memory awareness? Like it is aware of which file it interacted before, and so on?
No, I just made the game and that’s it
I just discovered Perplexity labs. Probably you already know about it.