Wafer

Overview

DeepIntShield connects to Wafer Serverless through its OpenAI-compatible API. Use the provider prefix wafer/. Provider-level operation gates remain fixed, while /v1/models supplies live model-level facts such as vision, tools, reasoning, structured output, Responses support, ZDR, context length, and pricing.

Supported operations

Operation	Non-streaming	Streaming	DeepIntShield route
List models and live metadata	✅	—	GET /v1/models
Text completions	✅	✅ SSE	POST /v1/completions
Chat Completions	✅	✅ SSE	POST /v1/chat/completions
Responses	✅	✅ SSE	POST /v1/responses
File upload	✅	—	POST /v1/files

File listing, retrieval, deletion, and content download are not enabled. Other unsupported operations fail locally before an upstream request.

Configure Wafer

1. Create a Wafer API key.
2. Open Workspace → AI Providers, select Wafer, and add the key.
3. Optionally restrict the key to an approved model allow-list.

The default upstream base URL is https://pass.wafer.ai. DeepIntShield uses a pooled client and bearer authentication. A custom network_config.base_url can target an approved private endpoint or proxy.

Make a streaming request

curl https://app.deepintshield.com/v1/chat/completions \
  -H "Authorization: Bearer $DEEPINTSHIELD_VIRTUAL_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "wafer/GLM-5.1",
    "messages": [{"role": "user", "content": "Propose three test cases."}],
    "stream": true
  }'

Before relying on vision, tools, reasoning, JSON Schema, regex/grammar output, Responses, or ZDR, inspect the selected model’s live metadata. A capability on one Wafer model does not imply the same capability on every Wafer model.

Upload a file

The upload surface accepts vision, document, or video purposes and rejects files larger than 512 MiB. For video, extraction_fps must be a finite positive number.

curl https://app.deepintshield.com/v1/files \
  -H "Authorization: Bearer $DEEPINTSHIELD_VIRTUAL_KEY" \
  -H "x-model-provider: wafer" \
  -F "purpose=document" \
  -F "file=@report.pdf"

Caution

This is upload-only support. Do not build a lifecycle that assumes the gateway can later list, retrieve, download, or delete the Wafer file.

Live pricing and discovery

GET /v1/models?provider=wafer converts Wafer’s live rates into DeepIntShield’s per-token prompt, completion, and cache-read pricing fields. It also preserves the context length, input modalities, supported methods, supported parameters, description, owner, and ZDR capability returned for each model. Inference does not call discovery, so model listing adds no lookup to the request path.

Use pricing overrides when your contract differs from the live public rate. Wafer adds models regularly; treat the live catalog as authoritative instead of hard-coding this page’s example.

See Wafer’s official serverless model catalog and API examples.