Skip to main content
Content parts are supplied under messages[].content, following the same shape OpenAI’s SDKs already use. Each part is one of: The VLM Run Gateway validates content parts against each model’s capabilities before inference. Mismatches return 400 with capability_violation. Check a model’s accepted input types on Models or GET /v1/openai/models.

​
Image Inputs

Most OCR and vision models accept a single image_url part per request. VQA models such as qwen/qwen3.5-0.8b accept multiple images in the same message, for side-by-side comparison or multi-image context.
image_url.detail (auto / low / high) is accepted for OpenAI SDK compatibility. The Gateway’s own documented cost/quality knob for images is image_resolution, below; prefer it when you want a specific resize behavior. Use image_resolution to resize an image before inference when a model’s default resolution costs more than the task needs. See Gateway extensions for the full parameter reference.

​
Video Inputs

qwen/qwen3.5-0.8b accepts one video_url part per request alongside an optional text prompt. The Gateway samples frames from the video before inference. Control how many frames reach the model with these top-level fields (or via extra_body in the OpenAI Python SDK):
Start with a low video_max_frames (for example 8) and tune video_fps before raising either knob. More frames increase latency and cost.

​
Document Inputs

Document (PDF) requests are decoded and rasterized at the ingress before per-page inference.

​
Limits

Requests over either limit return 400 with error.code = "invalid_document". See Error Codes.

​
URL vs. base64

Prefer a URL over base64 whenever possible, especially for documents over a few megabytes.
For document_dpi and other method-specific knobs, see Method Parameters.

​
Scanned vs. native PDFs

The VLM Run Gateway rasterizes every page and runs OCR regardless of whether the source PDF has a text layer. Scanned (image-only) and native (text-layer) PDFs are handled the same way, both go through the OCR model rather than a text-extraction shortcut. If you already have a reliable text layer and don’t need OCR, extracting it client-side before calling the Gateway will be faster and cheaper.

​
VQA on PDFs

qwen/qwen3.5-0.8b also accepts document_url PDFs, but unlike OCR models it does not run a dedicated OCR pipeline. The Gateway rasterizes each page and fans out a separate chat request per page, then concatenates the responses. Use this when you want free-form questions or summaries per page rather than structured markdown extraction. For OCR output, use a document model from Models instead.

Flexible Document OCR

End-to-end recipe from model selection to response parsing.

Streaming

Receive pages as they finish instead of waiting for the whole document.

Models

Which input types each model accepts.

Chat Completions

Full request and response schema.