Gemini via OpenAI Format — Files, Search, Thinking

Use Gemini models at POST /v1/chat/completions with the OpenAI SDK. Supports PDF, image, audio, video, web search grounding, and thinking budgets.

KiosAPI lets you call Gemini models using the standard OpenAI request format — pass files including PDFs, images, audio, and video directly in the content array of your messages. You can use the same OpenAI SDK and endpoint you use for other models; no additional configuration is required to switch to Gemini.

Supported File Types

Extension	MIME Type
.pdf	application/pdf
.mp3	audio/mp3
.mp4	video/mp4
.wav	audio/wav
.png	image/png
.jpg / .jpeg	image/jpeg
.txt	text/plain
.mov	video/mov
.mpeg	video/mpeg
.mpg	video/mpg
.avi	video/avi
.wmv	video/wmv
.flv	video/flv

Request

POST https://kiosapi.com/v1/chat/completions

Headers

Header	Required	Description
Content-Type	Yes	application/json
Authorization	Yes	Bearer sk-xxx

Request Body

model · string · required

Gemini model name (e.g. gemini-2.5-pro).

messages · array · required

Array of message objects forming the conversation.

Show messages properties

messages[].role · string · required

Role of the message author: user or assistant.

messages[].content · string | array · required

The message content. Pass a plain string for text-only messages, or an array of content objects for multimodal input.

max_tokens · number

Maximum number of tokens to generate in the response.

temperature · number

Sampling temperature.

stream · boolean

Set to true to stream the response using server-sent events.

tools · array

Function calling tools, or the special googleSearch tool for web search.

reasoning_effort · string

Thinking level for reasoning models. One of high, medium, or low.

extra_body · object

Advanced configuration, including thinking_config for fine-grained control over the thinking budget.

response_format · object

Structured output configuration.

Content Types for File Analysis

When passing content as an array, each element uses one of the following types:

Type	Fields	Description
text	text	Plain text prompt
file	file.filename, file.file_data	File passed as base64 data or a URL
file_url	file_url.url	File accessible by URL (images and files)
image_url	image_url.url	Image URL (base64 data URI or web URL)

For files over 20 MB, use the URL method rather than base64 to avoid hitting request size limits.

Dynamic Thinking Models

When a model name ends with thinking-* (e.g. gemini-2.5-flash-thinking-2000), it supports dynamic thinking token limits. The number after thinking- sets the maximum number of tokens the model may use for internal reasoning.

Web Search

You can enable web search in two ways:

1. Add the -search suffix to the model name (e.g. gemini-2.5-flash-search).
2. Include the googleSearch tool in your request body:

{
  "tools": [
    {
      "type": "function",
      "function": {"name": "googleSearch"}
    }
  ]
}

Examples

cURL (Chat)

curl https://kiosapi.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "model": "gemini-2.5-pro",
    "messages": [
      {
        "role": "user",
        "content": "Hello, who are you?"
      }
    ],
    "max_tokens": 1688,
    "temperature": 0.5,
    "stream": false
  }'

cURL (File Analysis)

curl https://kiosapi.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-xxx" \
  -d '{
    "model": "gemini-2.5-pro",
    "messages": [
      {
        "role": "user",
        "content": [
          {
            "type": "text",
            "text": "Summarize this document"
          },
          {
            "type": "file",
            "file": {
              "filename": "document.pdf",
              "file_data": "data:application/pdf;base64,JVBERi0xLjQK..."
            }
          }
        ]
      }
    ],
    "max_tokens": 6000,
    "stream": false
  }'

Python (Web Search)

import openai

openai.api_key = "sk-xxx"
openai.base_url = "https://kiosapi.com/v1/"

response = openai.chat.completions.create(
    model="gemini-2.5-flash-search",
    messages=[{"role": "user", "content": "What is today's date?"}],
    max_tokens=1688,
    temperature=0.5,
    stream=False
)

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

Python (Thinking Budget)

response = openai.chat.completions.create(
    model="gemini-2.5-flash",
    messages=[{"role": "user", "content": "Complex reasoning task..."}],
    max_tokens=8000,
    extra_body={
        "google": {
            "thinking_config": {
                "include_thoughts": True,
                "thinking_budget": 8192
            }
        }
    }
)

Response

{
  "id": "chatcmpl-xxx",
  "object": "chat.completion",
  "created": 1724972230,
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! I'm Gemini, a multilingual AI model..."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 38,
    "total_tokens": 50
  }
}