Nasiko announced TokenOps on November 16, 2026, a suite of spending-tracking and budget-control features built into its OpenRuntime platform. Operators can set a budget for each agent, receive an alert when 80% of the limit is reached, and automatically pause the agent when spending hits 100%. The platform centralizes token consumption and associated costs by agent, team, and model, including calls routed through providers such as OpenAI and Gemini. Nasiko estimates that effective model routing could reduce token spending by up to 50%, with no SDK migration or changes to existing agents and workflows required. OpenRuntime can be self-hosted or used as a hosted service and supports tools including Claude Code, Codex, Docker, OpenTelemetry, and an MCP gateway for managing permissions.
Source: Read the original article


