Add InferX provider (OpenAI-compatible endpoints)

[Prashanth-InferX]

Adds InferX as an OpenAI-compatible provider.

InferX provides dedicated inference endpoints with no cold starts.

Base URL format:
https://model.inferx.net/funccall/{tenant}/endpoints/{endpoint}/v1

Example model:
Qwen/Qwen3.6-27B-FP8

Compatible with OpenAI-style clients (e.g., OpenCode, AI SDK).

Users provide their own endpoint and API key.

[Prashanth-InferX]

Happy to add more models or adjust format if needed.

[Prashanth-InferX]

Hey there, just following up on the InferX provider PR from last week. Happy to make any changes needed if something needs to be adjusted for models.dev compatibility.

Would love to get InferX added as a supported provider when you have a chance. Thanks.

[Prashanth-InferX]

Hey Guys, really appreciate you taking a look at my request. Thank you/

[rekram1-node]

this isnt how u define models, so plz follow the guide

[Prashanth-InferX]

this isnt how u define models, so plz follow the guide

Hey, Thanks for the pointer and for taking a look. I followed the repo structure guide and opened a corrected PR here: #1843 . really appreciate it.

[Prashanth-InferX]

Thanks for reviewing. Just to clarify, was this closed because InferX uses endpoint-specific OpenAI-compatible base URLs, or is there another field/format issue I should fix?