Skip to main content
Wafer is a hosted inference provider for serverless chat models. TrueFoundry connects to Wafer natively through the AI Gateway, so you can send text prompts and receive text responses with optional zero-data-retention. Wafer supports chat completions, streaming, and tool / function calling (where the model supports it) for models like DeepSeek V4 Pro and DeepSeek V4 Flash.

​
Adding Models

This section explains the steps to add a Wafer provider account, register chat models, and configure access controls.
1

Navigate to Wafer in AI Gateway

From the TrueFoundry dashboard, navigate to AI Gateway > Models and select Wafer.
Wafer provider selection in TrueFoundry AI Gateway models page
2

Add Wafer account details

Click Add Wafer Account. Give a unique name to your Wafer account and complete the form:
  • Base URL (optional): defaults to https://pass.wafer.ai/v1. Change this only if Wafer directs you to a different endpoint.
  • Zero Data Retention: when enabled, the gateway sends the Wafer-ZDR: required header on API requests for request-scoped zero data retention. This is enabled by default.
  • API Key: your Wafer API key for authentication.
Add collaborators to your account to grant other users or teams access. Learn more about access control here.
Wafer account form with Base URL, Zero Data Retention, and API key fields
3

Register chat models

Click + Add Models to register one or more chat models under your Wafer account. For each model, set:
  • Display Name: the name shown in the TrueFoundry UI (for example, deepseek-v4-flash).
  • Model ID: the Wafer model identifier used in API calls (for example, deepseek-v4-flash or deepseek-v4-pro).
  • Model Types: select Chat.
Wafer model form with Display Name, Model ID, and Model Types
Use the exact Model ID from Wafer’s documentation or dashboard. Common models include deepseek-v4-flash and deepseek-v4-pro.

​
Inference

After adding the models, call them through the AI Gateway like any other chat model — via the Playground or by integrating with your application using the OpenAI-compatible /chat/completions API. Use the TrueFoundry model ID in the format your-wafer-account/deepseek-v4-flash. See Chat Completions - Getting Started for a full code example: 
from openai import OpenAI

client = OpenAI(
    api_key="your_truefoundry_api_key",
    base_url="{GATEWAY_BASE_URL}"
)

response = client.chat.completions.create(
    model="my-wafer-account/deepseek-v4-flash",  # provider account / model display name
    messages=[{"role": "user", "content": "Hello, how are you?"}]
)

print(response.choices[0].message.content)