# Gemini 4 Argon - Core.Today AI API

> Google's flagship reasoning model (GA 2026-10-13). A 1M-token context window with 65,536 max output tokens (thinking included), multimodal input, adjustable thinking levels and agentic tool use, priced at $4 input / $20 output per million tokens with implicitly cached inputs at $0.20/M.

- **Provider**: Google
- **Model ID**: gemini-4-argon
- **Category**: LLM
- **Credits**: 1 per request
- **Speed**: Medium
- **Quality**: High

## Model Specifications
- **Context Window**: 1.0M tokens
- **Max Output**: 66K tokens
- **Training Cutoff**: Not published
- **Supported Formats**: text, json, markdown
- **Compatible SDK**: OpenAI, Google AI

### Capabilities
- Vision (image input)
- Function Calling
- Streaming
- JSON Mode
- System Prompt

### Token Pricing (per 1M tokens)
- **Input Tokens**: 7,432 credits ($4.95)
- **Output Tokens**: 37,160 credits ($24.77)
- **Cached Tokens**: 371.6 credits ($0.25)

## Features
- Gemini 4 generation flagship reasoning model
- 1,048,576 token context window
- 65,536 max output tokens (thinking + output)
- Multimodal input: text, image, audio, video, PDF
- Thinking levels: low, medium (default), high
- Function calling, structured outputs, search grounding, code execution, URL context

## Use Cases
- Hard math, science and coding problems
- Agentic workflows with tools and search
- Long document and codebase analysis
- Multimodal reasoning over video, audio and PDFs

## API Endpoint

Base URL: https://api.core.today
Endpoint: POST /llm/gemini/v1beta/openai/chat/completions

## Authentication
Header: Authorization: Bearer YOUR_API_KEY

Note: LLM endpoints use OpenAI-compatible format with Authorization Bearer token.

## Input Parameters

### Required
- **messages**: array - Array of message objects (OpenAI format). Supports text, image, audio, video, and PDF inputs.

### Optional
- **max_tokens**: integer - Maximum output tokens, thinking included — a hard limit. Max: 65,536. Context window (input + output): 1,048,576 tokens.
- **response_format**: object - Output format constraint. Use `{ type: 'json_object' }` for structured JSON output.
- **stream**: boolean (default: false) - Enable Server-Sent Events streaming.

## Examples

### Hard Reasoning
A multi-step problem with room for thinking
```json
{
  "model": "gemini-4-argon",
  "messages": [
    {
      "role": "user",
      "content": "Prove that there are infinitely many primes of the form 4k+3."
    }
  ],
  "max_tokens": 32768
}
```

### Grounded Answer
Native Gemini API with Google Search grounding
```json
{
  "contents": [
    {
      "parts": [
        {
          "text": "What changed in the latest stable Go release?"
        }
      ]
    }
  ],
  "tools": [
    {
      "google_search": {}
    }
  ]
}
```

## Response Format
```json
{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Response text here"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 100,
    "completion_tokens": 50,
    "total_tokens": 150
  }
}
```

## Tips
- Thinking tokens are billed at the output rate and count toward max_tokens — raise it (32,768+) for hard tasks, or lower the thinking level to spend less
- temperature, top_p and top_k are ignored by this model
- Implicit caching only (prompts of 4,096+ tokens, cached input $0.20/M); explicit cachedContents is not supported
- Google Search grounding is billed at 26.012 credits ($14/1k) per search query the model issues, on top of tokens
- Priority inference (serviceTier: priority) bills token charges at 1.8x; flex and batch discounts are not passed through and settle at the standard rate
- Long decode (background: true, over 65,536 output tokens) is not supported through the gateway

## Documentation
https://ai.google.dev/gemini-api/docs/models/gemini-4-argon