feat: add ZenMux inference provider

[mingcheng]

• add ZenMux to list of supported providers in README
• implement ZenmuxConversationalTask for conversational AI
• update provider mappings and imports in getProviderHelper
• add zenmux to hardcoded model inference mapping
• include ZenMux in provider types and hub organizations

---

Note

Medium Risk
Adds a new external inference provider and wires it into provider selection/types, which may affect routing and header construction for conversational requests if misconfigured.

Overview
Adds ZenMux as a supported Inference Provider, including documentation updates and a new ZenmuxConversationalTask implementation.

Wires zenmux into provider resolution (getProviderHelper), hardcoded model mapping (HARDCODED_MODEL_INFERENCE_MAPPING), and public types/Hub org mappings so provider: "zenmux" can be used for conversational (chat completion) requests with ZenMux-specific request headers.

^{Reviewed by Cursor Bugbot for commit aab094e. Bugbot is set up for automated code reviews on this repo. Configure here.}

[mingcheng]

Hi, We are the ZenMux AI Team. We are AI FORCE SINGAPORE PTE. LTD, a team driven by technological innovation. Our vision is to fundamentally revolutionize the way AI applications are developed by eliminating infrastructure complexity and guaranteeing reliability.

ZenMux is our core platform, it is positioned as the world's first Enterprise-Grade Model Aggregation and Quality Assurance Solution that will deeply integrate AI Model Insurance Services. Our goal is to provide developers with a unified, stable environment that delivers the highest level of assurance for calling large models.

[mingcheng]

Hello, @julien-c.

Given that a week has passed, please let me know when this PR can be merged.

So many thanks

[cursor]

Cursor Bugbot has reviewed your changes 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 27f1632. Configure here.}

[cursor]

Duplicated v1 path segment in API URL

High Severity

The base URL "https://zenmux.ai/api/v1" already includes a /v1 segment, but the inherited BaseConversationalTask.makeRoute() returns "v1/chat/completions". When makeUrl combines them, the resulting URL becomes https://zenmux.ai/api/v1/v1/chat/completions with a duplicated v1. All direct API calls (with provider-key auth) will hit the wrong endpoint and fail.

 

^{Reviewed by Cursor Bugbot for commit 27f1632. Configure here.}