What
When load_test_mode is enabled, the gateway now writes to /dev/null and reads from /dev/zero for each connection, so the CPU used per server copy matches production. If the host has no /dev/null or /dev/zero, it logs a warning that CPU will read about a quarter to a third lower than in production, and the load test still runs.
Why
This only matters to operators measuring how much capacity the gateway needs, whose figures should now be more realistic.
Names in the bundleload_test_mode
The entry above is what we published on the day. These lines were added later, as Anthropic's own pages caught up, and they sit beside the original rather than replacing it.
Confirmed since
Anthropic's documentation has since written up load_test_mode, on Claude apps gateway configuration.
> Never turn this on for a gateway that developers use. Every request gets the canned reply and no model is called. The gateway logs a `load_test_mode is on` warning at boot and marks each `inference` [audit event](/docs/en/claude-apps-gat…claude-apps-gateway-config see the edit
Two sources agreeTwo things we can check say the same as this entry.
Anthropic's documentation agrees
Anthropic's documentation has since written up load_test_mode, on Claude apps gateway configuration.
Anthropic's release notes agree
Added an opt-in load_test_mode block to the Claude apps gateway config: requests are built and signed but not sent upstream, and clients get…