GLM 5.2 Nitro

GLM 5.2 Nitro is delivered through routing.run's model router, the same gateway documented on Mastra as exposing 15 models behind an OpenAI-compatible /chat/completions endpoint that authenticates with a ROUTING_RUN API KEY environment variable. Because applications reference it through routing.run's own model namespace rather than an upstream-provider identifier, client code stays uniform across the router's offerings while provider-specific behavior is delegated to the gateway. This router pattern lets teams already working with Mastra agents drop GLM 5.2 Nitro in without changing their request shape, relying on familiar streaming and generation ergonomics that come with OpenAI-compatibility.

In practical terms, GLM 5.2 Nitro functions as one of a set of comparable text-in, text-out endpoints reachable through a single API style, so selection between models in the routing.run catalog happens on the server side without rewriting callers. The Mastra framing positions it as a straightforward drop-in option for agent workflows that already use the router, while anyone evaluating it directly can lean on routing.run's own documentation for behavioral details that the gateway does not surface in the common request envelope.