feat(wafer/MiniMax-M3): add new models [bot]

[models-bot]

Auto-generated by model-addition-agent for wafer/MiniMax-M3.

---

Note

Low Risk
Adds a standalone model metadata file with no runtime or auth changes; typical low-risk catalog update from the model-addition agent.

Overview
Adds a new wafer provider definition for MiniMax-M3 via providers/wafer/MiniMax-M3.yaml.

The entry registers token pricing (including cache-read input), a 1M context window, function_calling, image input, and thinking: true, with mode: unknown.

^{Reviewed by Cursor Bugbot for commit 297d53b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

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

^{Reviewed by Cursor Bugbot for commit 297d53b. Configure here.}

[cursor]

Modalities omit text and video

Medium Severity

modalities.input lists only image, with no output block. MiniMax M3 is a multimodal chat model (text, image, and video in; text out). The same model in together-ai documents text, image, and video plus output: text, so this registry entry can misreport capabilities for routing, validation, or pricing.

 

^{Reviewed by Cursor Bugbot for commit 297d53b. Configure here.}

[cursor]

Chat model uses unknown mode

Medium Severity

mode is set to unknown even though this entry describes a chat-capable MiniMax M3 model with function_calling. Peer registry entries for the same model use mode: chat and supportedModes: [chat], so unknown can hide the model from chat routing and tooling that filters on mode.

 

^{Reviewed by Cursor Bugbot for commit 297d53b. Configure here.}