llama-dash is a self-hosted inference gateway providing a single dashboard for monitoring models, requests, API keys, routing rules, and proxy metrics.

It fronts llama-swap and OpenAI- or Anthropic-compatible upstreams, with client support for OpenAI SDK, Claude Code, Continue, and Open WebUI. Features include model load/unload with config validation, searchable request history with cost estimates, per-key rate limits, routing rules, and model playgrounds.