Build with term-llm - SambaNova Integration Guide

term-llm is an open-source, terminal-first AI runtime written in Go. It turns natural language into executable shell commands, runs persistent chats with tools and MCP servers, edits files with model assistance, and supports agents, skills, sessions, jobs, and local automation. Follow the steps below to connect term-llm with a model hosted on SambaCloud.

Prerequisites

Before you begin, ensure you have:

A SambaCloud account and API key
A POSIX shell on macOS, Linux, or WSL
Go 1.22+ (only if you plan to build from source)

Setup

Install term-llm

Install term-llm using the official installation script:

curl -fsSL https://raw.githubusercontent.com/samsaffron/term-llm/main/install.sh | sh

Or install it with Go:

go install github.com/samsaffron/term-llm@latest

If term-llm --version returns command not found (common on macOS), add the install directory to your PATH:

export PATH="$HOME/.local/bin:$PATH"

To persist this change, append the same line to ~/.zshrc (default on macOS) or ~/.bashrc, then restart your terminal or run source ~/.zshrc.

Configure your SambaNova API key

Export your SambaCloud API key:

export SAMBANOVA_API_KEY=<your-api-key>

Alternatively, you can configure the API key directly in ~/.config/term-llm/config.yaml under providers.sambanova.api_key.

Configure SambaNova as the default provider

Set SambaNova as your default provider and configure its models:

term-llm config set default_provider sambanova
term-llm config set providers.sambanova.model gpt-oss-120b
term-llm config set providers.sambanova.fast_model Meta-Llama-3.3-70B-Instruct

These commands create ~/.config/term-llm/config.yaml if it doesn’t exist yet, and edit it in place. Tab-complete a model ID with term-llm config set providers.sambanova.model <TAB>.Or edit ~/.config/term-llm/config.yaml directly:

default_provider: sambanova

providers:
  sambanova:
    model: gpt-oss-120b
    fast_model: Meta-Llama-3.3-70B-Instruct

Start using SambaNova models

Run prompts using your configured provider.Ask a question:

term-llm ask "Explain RDUs"

Translate natural language into a shell command:

term-llm exec "Find all Go files modified today"

Start an interactive session:

term-llm chat

To override the configured model for a single request:

term-llm ask \
  --provider sambanova:Meta-Llama-3.3-70B-Instruct \
  "Summarize this repository"

Available models

List all chat-capable SambaNova models along with their context windows and pricing information:

term-llm models --provider sambanova

term-llm includes a bundled pricing table synchronized with SambaNova’s public pricing information.

Video walkthrough

Follow the video below to install term-llm, configure SambaNova, and start using SambaCloud models.

Troubleshooting

If term-llm isn’t found after installation (common on macOS), add the install directory to your PATH:

export PATH="$HOME/.local/bin:$PATH"

To persist this change, append the same line to ~/.zshrc (default on macOS) or ~/.bashrc, then restart your terminal or run source ~/.zshrc. If SAMBANOVA_API_KEY isn’t being recognized, verify that it’s set and doesn’t include the literal angle brackets:

echo "${SAMBANOVA_API_KEY:0:10}..."

If the output is empty or incorrect, re-export your API key or generate a new one in the SambaCloud portal. If requests fail because the configured model isn’t available, list the current SambaNova models:

term-llm models --provider sambanova

Then update the model: field in ~/.config/term-llm/config.yaml to one of the listed model IDs. If ~/.config/term-llm/config.yaml doesn’t exist, create the configuration directory first:

mkdir -p ~/.config/term-llm

Overview

Developer tools

Frameworks

Interfaces and low-code

Data and infrastructure

Multimodal

Evaluation and monitoring

term-llm integration guide

Prerequisites

Setup

Available models

Video walkthrough

Troubleshooting

Additional resources

​Prerequisites

​Setup

​Available models

​Video walkthrough

​Troubleshooting

​Additional resources

Prerequisites

Setup

Available models

Video walkthrough

Troubleshooting

Additional resources