feat(openrouter/z-ai/glm-5.3): add new models [bot]

[models-bot]

Auto-generated by model-addition-agent for openrouter/z-ai/glm-5.3.

---

Note

Low Risk
Adds static model metadata only; no runtime, auth, or billing logic changes.

Overview
Adds a new OpenRouter provider catalog file for z-ai/glm-5.3, following the same pattern as other GLM 5.x entries.

The entry defines chat mode with thinking enabled, text-only I/O, serverless provisioning, and defaults for temperature (1) and top_p (0.95). Compared to glm-5.2, it advertises a larger context window (1,310,720 vs 1,048,576), the same input/output token pricing with a lower cache-read rate, and an extra capability flag parallel_function_calling alongside the usual tool/JSON/structured-output features.

Pricing and limits are sourced from the OpenRouter model page.

^{Reviewed by Cursor Bugbot for commit 470c02b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit a67b8b7. Configure here.}

[cursor]

Missing reasoning_effort param override

Medium Severity

This thinking GLM-5.3 entry inherits OpenRouter's reasoning_effort default of medium, which the model does not accept. OpenRouter and Z.ai document low, high, and max (default max). Requests that send the inherited default can fail or run at the wrong reasoning depth.

 

^{Triggered by learned rule: Override inherited reasoning_effort when default isn't supported}

^{Reviewed by Cursor Bugbot for commit a67b8b7. Configure here.}

[github-actions]

/test-models