Run Nemotron 3 Ultra in Your Terminal ⚡
NVIDIA's 550B open reasoning model, wired into OpenCode as your autonomous terminal coding agent. No GPU. No subscription. Copy, paste, done.
You are not downloading a 550B model onto your laptop (which would require ~4× NVIDIA B200 GPUs). Instead, NVIDIA hosts the model on their cloud infrastructure and provides free API access, while OpenCode runs locally as a lightweight CLI agent in your terminal. Your API key bridges the two.
Nemotron 3 Ultra uses a Hybrid Mamba-Transformer Mixture-of-Experts (MoE) architecture with a massive 1,000,000 token context window. It's built specifically for long-running agentic tasks: planning, deep repository refactoring, and multi-file error self-correction.
Install OpenCode CLI
Select the installation method for your operating system:
winget install SST.opencode
npm install -g opencode-ai@latest
scoop bucket add extras; scoop install extras/opencode
cmd.exe for full ANSI color and interactive terminal rendering.
curl -fsSL https://opencode.ai/install.sh | bash
Verify Installation
Close your terminal and open a brand new window so PATH variables refresh. Run:
opencode --version
If PowerShell gives an execution policy error on Windows, run this one-time fix: Set-ExecutionPolicy -Scope CurrentUser RemoteSigned
Get Your Free NVIDIA API Key
- Visit build.nvidia.com/settings/api-keys or explore the Nemotron 3 Ultra Model Page.
- Sign in or sign up with your personal email (e.g. Gmail).
- Navigate to Settings → API Keys → Generate API Key.
- Copy the key immediately (starts with
nvapi-). It is displayed only once.
Connect NVIDIA & Launch Nemotron
Move into any project directory and launch OpenCode:
cd $HOME/Desktop/my-project opencode
Inside the interactive OpenCode terminal screen, run the following slash commands:
/connectSelect NVIDIA from the list, paste your
nvapi- key, and press Enter.
/model nvidia/nemotron-3-ultra-550b-a55bNow ask it: "Read this codebase and explain what the system does." It will read files, write code, run commands, and debug errors autonomously.
Direct Python Scripting (OpenAI-Compatible)
If you want to call Nemotron 3 Ultra directly in your Python pipelines, use the OpenAI client with NVIDIA's base URL:
import os
from openai import OpenAI
client = OpenAI(
base_url="https://integrate.api.nvidia.com/v1",
api_key=os.environ["NVIDIA_API_KEY"],
)
stream = client.chat.completions.create(
model="nvidia/nemotron-3-ultra-550b-a55b",
messages=[
{"role": "user", "content": "Write a Python script that renames every file in a folder to lowercase."}
],
temperature=1,
top_p=0.95,
max_tokens=16384,
extra_body={"chat_template_kwargs": {"enable_thinking": True}},
stream=True,
)
for chunk in stream:
if chunk.choices and chunk.choices[0].delta.content:
print(chunk.choices[0].delta.content, end="")
Common Issues & Instant Fixes
| SYMPTOM | RESOLUTION |
|---|---|
| 'opencode' not recognized | Open a fresh terminal window to reload PATH. If still failing, install via npm. |
| running scripts is disabled | Run Set-ExecutionPolicy -Scope CurrentUser RemoteSigned in PowerShell. |
| 401 Unauthorized | API key is invalid, expired, or has a leading/trailing space. Generate a fresh key. |
| No API Key Permissions | Wrong organisation selected. Click Switch Org → personal default org. |
| Nemotron not in /models | Provider is not connected. Run /connect again and select NVIDIA. |
| Replies feel deliberate / slow | Nemotron 3 Ultra is an open reasoning model—it outputs chain-of-thought before code. |