Gateway for coding agents that compresses tokens before requests reach LLM providers such as Anthropic and OpenAI, cutting cost by up to 50%. Also routes to cheaper or faster models and falls back automatically during provider outages or plan limits.