Ampersand from Humans& now available on AI Gateway

Ampersand, a language model from Humans&, is now accessible on Vercel’s AI Gateway at no cost through November 13, 2026. The model is engineered to engage in group conversations by deciding whether to speak or remain silent based on the discussion’s context and goals. It can ask clarifying questions, share relevant information, or wait for others to respond, using caller-defined tools like send_message and pass to execute actions. Developers must avoid required or named tool choices, as Ampersand only accepts auto or none.

Integration with AI Gateway provides a unified API for calling models, tracking usage and costs, and configuring retries, failover, and performance optimizations. The platform offers built-in custom reporting, API key budgets, routing rules, and higher uptime than individual providers. Pricing mirrors provider rates with no platform markup, including for Bring Your Own Key requests.

To use Ampersand in coding agents like Claude Code, Codex, or Cursor, install the latest Vercel CLI, run setup, and select humansand/ampersand-v0.1 in the agent. The model operates via the AI SDK, OpenAI-compatible Chat Completions API, or coding agents, with tool calls managed through the application.

For deployed chat applications, developers should persist responses, conversation history, and delivery receipts in durable storage. Messages must be processed sequentially, with retries handled using saved actions and receipts, and access checks performed before posting contributions. Handling pass records waiting actions without generating new model turns or messages.