OpenCode + gpt-oss

(Back to the choices.)

gpt-oss needs 16 GB of memory and about 20 GB of free disk space.

Install Ollama, gpt-oss, and OpenCode. Paste the block into PowerShell (Windows) or Terminal (Mac). The model download takes a while.

On Windows:

irm https://ollama.com/install.ps1 | iex
ollama pull gpt-oss:20b
winget install SST.opencode --accept-source-agreements --accept-package-agreements
New-Item -ItemType Directory -Force "$HOME\.config\opencode" | Out-Null
@'
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "ollama": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Ollama (local)",
      "options": { "baseURL": "http://localhost:11434/v1" },
      "models": {
        "gpt-oss:20b": { "name": "gpt-oss 20B (local)" }
      }
    }
  }
}
'@ | Set-Content "$HOME\.config\opencode\opencode.json"

On Mac:

curl -fsSL https://ollama.com/install.sh | sh
ollama pull gpt-oss:20b
curl -fsSL https://opencode.ai/install | bash
mkdir -p ~/.config/opencode
cat > ~/.config/opencode/opencode.json <<'EOF'
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "ollama": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Ollama (local)",
      "options": { "baseURL": "http://localhost:11434/v1" },
      "models": {
        "gpt-oss:20b": { "name": "gpt-oss 20B (local)" }
      }
    }
  }
}
EOF

(It may ask for your Mac password; the typing doesn’t show.)

Set the context length. Click the Ollama icon (in the menu bar on a Mac, or the system tray on Windows) > Settings, and move the Context length slider to 64k.

Check. Open a new terminal window and run

opencode run --model ollama/gpt-oss:20b "Reply with only the word hello."

It should print hello. (It can take a minute or two.)

Then run ollama ps and make sure the CONTEXT column says 65536. Then go to A First Example.

If you get stuck, copy the command and everything it printed and send it to Bryson or me.