Ollama

“Been using qwen3.5:397b-cloud for three weeks and it has done some meaty research, analyis and production tasks. It's also made a god job of OC configuation and and bug fixing tasks. Giving GLM 5.1 a spin after seeing a chap benchmarch 24 models with his existing (daily health monitoring) tasks and it pipped Qwen to the post. :) Only flipped it this morning, so time will tell.”

“btw u should use llama.cpp instead of ollama”

“The model management is nice and the fact that they make sure the instruct prompt format is always correct is KEY imo I no longer use llama.cpp server in my evals because of the chat format issue specifically, without Jinja support it's very hit and miss as soon as you step off the baked in templates.”

“Compile llama.cpp on your hardware and delete Ollama and double your inference speed .”

“Disclaimer: I am not affiliated with Ollama in any way. This is purely based on my personal experience as a long-term user. I’ve been using Ollama since it first launched, and it has genuinely changed my workflow. Even with a powerful local machine, there are certain walls you eventually hit. Lately, I’ve been testing the $20/month Cloud plan, and I wanted to share why I think it’s worth every penny. The "Large Model" BarrierWe are seeing incredible models being released, like Kimi-k2.5, DeepSee”

“Most of my Hermes agents are routed through Opencode Go. I barely used Ollama Cloud and I'm already at my weekly limit. It's disappointing that we got rug-pulled without any communication or warning. Based on my usage, the Deepseek API is approximately the same token value as Ollama Pro...”

“hey i have the ollama cloud i get it for first it get a 4 days and 4 hours rest period after 5 days the it gets 6 days reset period . i ahev the max 100$ plan . very disappointed . Such a scam and gaslighting product . don't ever get it . i also suggest . don't even get the local ones . they are just gaslighting company .”

“I have a huge data set that grows every day, I use ollama and a couple of computers with nvidia graphics cards to pull and process that data each day. it builds summaries, finds irregularities, and patterns very easily. These were computers we had in the office that went being used much. Sending that data to ChatGPT would have cost about 5-8.00 a day and shared information I really didnt want to share with OpenAI or anyone else. its simple and does all we need.”

“Because Ollama just works. I always run into Python dependency issues with textgen webui, and llamacpp has nowhere near the level of plug-and-play usability. I've never had any problems with Ollama besides CORS issues which was an easy fix. It's simple, well-documented, and just works. I never have to worry about prompt formats and pulling or running models is ridiculously easy.”

“hey i have the ollama cloud i get it for first it get a 4 days and 4 hours rest period after 5 days the it gets 6 days reset period . i ahev the max 100$ plan . very disappointed . Such a scam and gaslighting product . don't ever get it . i also suggest . don't even get the local ones . they are just gaslighting company .”

“I came across this article the other month and saw that there was (maybe still is) a potential vulnerability in running ollama with a Docker deployment. https://medium.com/@cyberprosocial/vulnerability-in-ollama-ai-platform-raises-remote-code-execution-concerns-d30adbcfbc1a This made me wonder if there's any potential security risk in running Ollama on localhost and only using well-known, open-source models like some of the ones here: https://ollama.com/library. Due to privacy, I mainly want to ”

“DNS blocked everywhere.”

“Ollama is ridiculous simple to install and test.Two commands you have a model running.It's limited in configuration... but when you are starting, the lamma cpp options are completely overwhelming.Even the normal stuff like temperature is sketchy to understand when you just started.”

“First of all... I downloaded I downloaded the windows application of ollama, I ain't setting up any thing in cmd because I hate commands... And after downloading it, i downloaded few models using huggingface commands... It downloaded and when I wanted to run, it don't show models... Not even any commands work... Like when I type a command it just gives me LLM generated response and I hate commands!! It's literally made for linux people cause i ain't pulling off so much commands just to use this ”

“I like that it automatically handles so much Want a new model? Pull it Want to test an application that's calling the API endpoints? It loads it Finish that work? It unloads it Apart from a couple of models I've added manually it saves me thinking about it and I can develop with it almost as seamlessly as using Open API”

“Uninstalled textgen webui today because I got Ollama and OpenWebUI. Perfect combo because it's let's you search web pages and upload your own documents as context. For me the biggest selling point of Ollama is the ease of installation”