Skip to content

Back to the catalog

Akash Rajagopal (@akashsairaja) on X
x.comthread

Akash Rajagopal (@akashsairaja) on X

Ollama overview thread by Akash Rajagopal on X. It lists six steps: install on macOS, Linux, Windows or Docker; run `ollama run gemma4`; use the REST API on localhost:11434 (/api/chat); call it from ollama-python and ollama-js; llama.cpp as the README's supported backend; and `ollama launch claude` to connect Claude Code. Codex, OpenCode, Copilot CLI and Droid also work. The thread cites 184k GitHub stars and an MIT licence.

TL;DR Thread explains how Ollama runs local models, exposes a REST API, and launches Claude Code, Codex and OpenCode against open models with one command.
Open original ↗
Document

What it claims

The thread by Akash Rajagopal describes Ollama, which runs open models locally and connects coding agents to them. It cites 184k GitHub stars and an MIT licence. Two details are not in the catalog summary. The Docker install uses the ollama/ollama image. Running ollama with no arguments prompts the user to pick a model or an agent.

How to use it

Install on macOS, Linux or Windows, or pull the ollama/ollama Docker image. Start an interactive picker for a model or agent:

ollama

Pull gemma4 from Ollama's library and open a chat:

ollama run gemma4

Connect Claude Code to Ollama's models:

ollama launch claude

The thread says Codex, OpenCode, Copilot CLI and Droid also work, but gives no launch command for them. The server that runs models also serves a REST API on localhost:11434, and the thread names only the /api/chat endpoint. The ollama-python and ollama-js clients call the same server from code.

Limits

The thread is a short promotional overview. It gives no model sizes, hardware or memory requirements, quantisation options, request or response schema for /api/chat, or speed figures. It does not say how well the coding agents perform against open models. The statement that llama.cpp is the supported backend is attributed to the README, not shown. The star count and licence are the thread's claims as of its posting and should be checked against the GitHub repository before relying on them.

Organisations named here

  • Ollama
  • Akash Rajagopal
  • Claude Code
  • GitHub Copilot
  • OpenCode

Technologies named here

  • Gemma 4
  • Ollama
  • ollama launch
  • ollama-python
  • ollama-js
  • llama.cpp
  • Claude Code
  • Codex
  • OpenCode
  • Copilot CLI
  • Droid
  • Docker
  • GitHub Copilot
  • Gemma
  • JavaScript
  • Linux
  • macOS
  • Python
  • Windows

Techniques named here

  • Running coding agents on open models via Ollama

Related in inference