feat(wafer/Qwen3.6-35B-A3B): add new models [bot]

[models-bot]

Auto-generated by model-addition-agent for wafer/Qwen3.6-35B-A3B.

---

Note

Low Risk
Single declarative catalog file with no application or routing logic changes; verify pricing and capability flags match wafer’s actual offering.

Overview
Adds a new wafer provider model definition for Qwen3.6-35B-A3B, registering token pricing (including cache-read input), a 256k context window, function_calling, thinking, and image input modality. mode is set to unknown, consistent with an agent-generated stub pending fuller metadata.

^{Reviewed by Cursor Bugbot for commit 5b2852b. 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 5b2852b. Configure here.}

[cursor]

Chat model uses unknown mode

High Severity

The registry entry sets mode to unknown even though the model declares function_calling and thinking, which match a chat/reasoning API. Peer Qwen and other tool-capable models in this repo use mode: chat, and the schema docs treat mode as the primary capability. Downstream consumers of dist/ai-models.json may not route or expose this model through chat endpoints.

 

^{Reviewed by Cursor Bugbot for commit 5b2852b. Configure here.}

[cursor]

Input modalities omit text

Medium Severity

modalities.input lists only image, with no text (or output text). Qwen3.6-35B-A3B is a general chat model; the sibling DeepInfra registry entry includes text, image, and video. Advertising image-only input misstates what callers can send and can break capability checks that rely on modalities rather than inferred chat mode.

 

^{Reviewed by Cursor Bugbot for commit 5b2852b. Configure here.}