Ollama Cloud vs your own server: what changes
Ollama Cloud and a self-hosted Ollama share one CLI and one API. See exactly which settings change, and what leaves your machine on each path.
#ollama
Filtering by topic #ollama · clear
Import a GGUF model into Ollama
Run a .gguf from Hugging Face or from a local file under Ollama, and fix the chat template mismatch that makes an imported model reply in garbage.
The Ollama API and port 11434 explained
Curl your Ollama server on port 11434, read what a refused connection means, tour the API endpoints, and set OLLAMA_HOST without exposing the box.
Ollama concurrency: NUM_PARALLEL and MAX_QUEUE
The second request waits, or it is refused: how OLLAMA_NUM_PARALLEL and OLLAMA_MAX_QUEUE decide which, and why every parallel slot costs you VRAM.
Reasoning effort settings on a local LLM
Reasoning effort sets how long a local model thinks before it answers. What the levels change on your own hardware, and how to measure the real cost.
Configure dsh: API keys, models, endpoints
Where dsh keeps its config on Linux, how to wire a DeepSeek API key or a local Ollama endpoint, and exactly what leaves your box in each mode.
Run GLM 5.2 on a VPS with Ollama
GLM 5.2 is cloud only in Ollama's library. Here is the GLM model that actually fits a VPS, and the RAM each quantisation needs on your own box.
Run DeepSeek V4 Flash on a VPS
DeepSeek V4 Flash on Ollama is cloud only. Get the real disk and RAM numbers for every GGUF build, and see which VPS plan can actually load it.
Run Meta Muse Glimmer 30B on a VPS
Muse Glimmer tags run from 17GB to 59GB. Work out the RAM and disk a rented Linux VPS needs before you pull, and what CPU only inference costs.
Ollama pull vs run, and where models live
ollama pull downloads a model and stops. ollama run downloads it, then opens a chat. Where the files land, why they fill a VPS root disk, how to move them.
Ollama num_predict: cap the output length
num_predict caps how many tokens Ollama will write. The three places to set it, which one wins, and how to read done_reason in the response.
Your Ollama API has no password
The Ollama server ships with no authentication, so anything that reaches port 11434 can run your models and pull new ones. The three fixes, in order.
Run Ollama in rootless Podman on a VPS
Run Ollama as a rootless Podman container on a VPS: a dedicated user, lingering, a Quadlet unit that survives reboot, SELinux labels, and a closed port.
Run Nemotron 3.5 Lightning on a VPS
Run NVIDIA Nemotron 3.5 Lightning with Ollama on your own server: the exact tag to pull, how much RAM it needs, and if CPU-only is fast enough.
Ollama quantization: Q4 vs Q8 vs fp16
Pick an Ollama quantization with arithmetic instead of guesswork: what q4_K_M, q8_0 and fp16 cost in RAM, and where the quality actually drops.
Which AI models can you self-host?
Pick a model by the RAM you actually have. Sizing arithmetic for 4 GB, 16 GB and 64 GB VPS plans, honest CPU token rates, and the hidden cost of context.
Ollama context length: setting num_ctx
Ollama truncates long prompts at a small default context window. Set num_ctx per request or server wide, and size the KV cache RAM before you raise it.
Keep an Ollama model loaded in memory
Ollama unloads your model after 5 minutes idle, so the next request pays the full load time again. Set keep_alive so it sticks, even after a reboot.
Why your self-hosted LLM stalls at 5 users
One user was fine, five crawl. How batching, KV cache limits, prefill and queue depth decide how many people your LLM server can serve at once.
Use Ollama with your coding agent
Point a coding agent at a model you host: the base URL, the dummy key, the context length that breaks everything, and the jobs a local model wins.
Self-host Octop: a multi-user AI assistant
Deploy Octop on a VPS with Docker Compose pinned to a tag: per-user isolation, an OpenAI-compatible model backend, TLS, and why to skip the curl installer.
Measure tokens per second on a local LLM
A rented GPU only beats per-token billing above a throughput floor. Measure tokens per second properly with a concurrency sweep, then decide.
Run Qwen 3.8 27B on a VPS With Ollama
There is no Qwen 3.8 on Ollama yet. Here is the arithmetic for running the 27B tag that does exist on a CPU-only VPS, and what fits in 8 to 64 GB.
Self-host mem0 for agent memory on a VPS
Run the mem0 memory server on your own VPS: the real RAM floor, a Compose file that binds to localhost, TLS in front of the API, and a fully local Ollama path.
Open WebUI alternatives for a VPS
Open WebUI, LibreChat, Hollama and OrionChat compared on a VPS with a public IP: RAM left for the model, logins, remote Ollama, and upkeep.
Ollama vs llama.cpp on a VPS
llama.cpp is the engine, Ollama is the layer on top. Which one to run on a CPU-only VPS, how quantisation choice changes RAM, and when neither fits.
Ollama vs vLLM: which LLM server to run
Ollama is a convenience layer for one user, on CPU if needed. vLLM is a throughput engine for GPUs. Pick by workload, with the real commands for both.
Can you self-host Claude? The honest answer
Claude's weights are not public, so no server of yours can run it. Here is what you can self-host instead: open models, a gateway, and Claude Code.
Host Ollama on a VPS: self-host an LLM safely
A 7B model wants 8 GB of RAM and gives 4 to 10 tokens per second on CPU. Host one on a VPS and call it at 127.0.0.1:11434/v1, with port 11434 kept closed.
Run a coding AI agent on a VPS
Run Aider or Goose on a VPS to code from your terminal, connected to a hosted or a self-hosted Ollama model. Setup, model sizing, and safety.