@hiway2llm/cli

HiWay2LLM CLI — smart LLM routing with BYOK, from your terminal.

Save 40–60 % on your LLM bill by routing every request to the cheapest capable model. You keep your own Anthropic / OpenAI / Google / Mistral / DeepSeek keys — HiWay never marks up inference.

Install

npm install -g @hiway2llm/cli

Cross-platform (Windows, macOS, Linux). Requires Node.js 18.17 or newer. No binary download, no SmartScreen prompt.

Quick start

# 1. Sign up (30 seconds, no credit card)
hw signup

# 2. Mint an API key (after email verification)
hw keys create "laptop"

# 3. First routed request
hw chat "Explain the pigeonhole principle in 3 bullets"

The first key you mint is cached as the default for hw chat. You can override it in ~/.hiway/config.json.

Commands

| Command | What it does | |---------------------------------|-------------------------------------------------------| | hw signup | Create an account (interactive) | | hw login | Sign in to an existing account | | hw logout | Wipe local credentials | | hw whoami | Show current user + plan | | hw keys list | List your HiWay API keys | | hw keys create <name> | Mint a new hw_live_… key | | hw keys revoke <id> | Revoke a key | | hw providers list | List your BYOK provider keys and their status | | hw chat "<prompt>" | Single-turn chat via smart routing (model=auto) | | hw chat -m <model> "<prompt>" | Force a specific model (e.g. openai/gpt-4o) | | hw chat --json "<prompt>" | Print the raw API response |

Config

~/.hiway/config.json     # chmod 600 — holds session token + api_key
HIWAY_API_URL            # override the API base (default: https://app.hiway2llm.com)

See also

• Python SDK — pip install hiway2llm
• TypeScript SDK — npm install @hiway2llm/client
• OpenAI-compatible API — point any OpenAI SDK at https://app.hiway2llm.com/v1 with your hw_live_… key
• Dashboard — app.hiway2llm.com — BYOK vault, Guardian anti-loop, Budget Control, A/B experiments, semantic cache, PII masking

License

MIT — © Mytm-Group SAS