Ollama launch: One‑Command Tool Setup and New 5‑Hour Cloud Sessions

ollama launch: single‑command evolution

Ollama introduced the ollama launch command, which configures and starts a selected programming tool without requiring environment variables or configuration files.

Supported programming tools

Claude Code

OpenCode

Codex

Droid

Previous manual setup for OpenCode

Download the Ollama client and start a model.

Create an opencode.json configuration file.

Write provider settings (e.g., baseURL, models).

Add the provider manually in the OpenCode client.

Enter the API key.

Quick start

Step 1: Install Ollama v0.15+ – download the latest version from ollama.com/download.

Step 2: Pull a programming model

# Local model (≈23 GB VRAM, 64 000‑token context)
ollama pull glm-4.7-flash
# Cloud model with full context length
ollama pull glm-4.7:cloud

Step 3: Launch a tool

ollama launch claude
ollama launch opencode

The system prompts you to select a model and then starts the integrated tool.

Recommended programming models

Local models glm-4.7-flash – fast, strong coding ability from the GLM family. qwen3-coder – Alibaba’s coding‑focused Qwen version. gpt-oss:20b – open‑source GPT‑style 20‑billion‑parameter model.

Cloud models glm-4.7:cloud – cloud version of GLM‑4.7 with full context. minimax-m2.1:cloud – domestic strong model from MiniMax. gpt-oss:120b-cloud – 120 B‑parameter model. qwen3-coder:480b-cloud – 480 B coding‑focused Qwen3, the strongest coding model.

Important tip: For programming tools, set the context length to at least 64 000 tokens in Ollama’s settings.

Configure without launching

Run ollama launch opencode --config to complete all configuration steps without starting the application; subsequent ollama launch opencode will start directly.

Cloud programming sessions

If local hardware cannot handle the models (e.g., 23 GB VRAM), Ollama provides cloud models. The latest update extends the free programming session window to five hours.

Ollama evolution roadmap

Standardized local deployment : ollama run runs a model with a single command.

Cloud model integration : ollama pull xxx:cloud pulls a cloud model when local resources are insufficient.

Tool integration standardization : ollama launch runs a tool with a single command.