feat(deepinfra): update model YAMLs [bot]

[hganwani-droid]

Auto-generated by poc-agent for provider deepinfra.

---

Note

Medium Risk
Updates provider model metadata (modalities, vector sizes, token limits, and supported features), which could affect model selection/validation and max-token enforcement if any values are incorrect.

Overview
DeepInfra model configs are refreshed to more completely describe capabilities and constraints.

Embedding models gain output_vector_size and more consistent modalities (primarily text input), while many image models add explicit modalities.output: image and/or modalities.input: text where applicable.

Several chat models update declared features (e.g., function_calling, structured_output, tools) and adjust limits fields (notably max_tokens/max_output_tokens), plus a few source URL corrections/additions.

^{Written by Cursor Bugbot for commit 8cdf11a. This will update automatically on new commits. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes and found 3 potential issues.

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

[cursor]

FLUX-2-dev missing input modality after key rename

High Severity

The input key under modalities was replaced with output instead of adding output alongside input. The model now has no input modalities defined at all. Every other FLUX-2 sibling model (FLUX-2-pro, FLUX-2-max, FLUX-2-klein-4b) and the similar FLUX-1-dev all define input modalities (e.g., text). This will likely cause consumers to believe the model accepts no input, breaking model selection or validation logic.

 

[cursor]

MiniMax-M2.5 lost max_output_tokens and max_tokens limits

High Severity

The max_output_tokens and max_tokens fields were removed from the limits section, leaving only context_window: 196608. The sibling model MiniMax-M2.1 retains both max_output_tokens: 131072 and max_tokens: 131072. This data loss means consumers of this config have no information about the model's output token limits, which could lead to incorrect request sizing or validation failures.

 

[cursor]

Seed-2.0-mini lost max_tokens without adding max_output_tokens

Medium Severity

The max_tokens: 256000 field was removed from limits without adding a max_output_tokens value, leaving only context_window: 256000. The sibling model ByteDance/Seed-1.8 retains both max_output_tokens: 256000 and max_tokens: 256000. This data loss removes output token limit information for this model.