# GPT-5.4 mini

> OpenAI's compact model for coding, computer use and subagents, with tool calling and image input.

## Key facts

- Model author: OpenAI
- Input: $0.75 per 1M tokens
- Output: $4.50 per 1M tokens
- Cached input: $0.075 per 1M tokens
- Context window: 400,000 tokens
- Maximum output: 128,000 tokens
- Input modalities: image, text
- Output modalities: text
- Capabilities: apply_patch, code_interpreter, computer_use, file_search, hosted_shell, image_generation, reasoning, structured_output, tools, vision, web_search
- Released: 2026-03-17
- Knowledge cutoff: Aug 31, 2025
- Model IDs: `openai/gpt-5.4-mini`
- Model page: https://www.realrelay.ai/model/openai/gpt-5.4-mini

## Overview

GPT-5.4 mini is a compact model from OpenAI designed for coding, computer use and subagents. It accepts text and image input, returns text, and supports reasoning, tool calling and structured output. Its mini size suits workflows that make many model calls.

It has a 400K-token context window and can generate up to 128,000 output tokens. It accepts text and image input and returns text. Its knowledge cutoff is August 31, 2025, and it was released on March 17, 2026.

## Capabilities

GPT-5.4 mini supports reasoning, tool calling, structured output and image input, and returns text output. Its listed tools also include web search, file search, a code interpreter, computer use, image generation, patch-based code editing and a hosted shell.

OpenAI documents GPT-5.4 mini for coding, computer use and subagents.

## Pricing

GPT-5.4 mini costs $0.75 per 1M input tokens and $4.50 per 1M output tokens. Cached input reads cost $0.075 per 1M tokens. GPT-5.4 mini is available through RealRelay's OpenAI-compatible API.

## Use cases

Use GPT-5.4 mini as a subagent: an orchestrating agent can hand it focused coding or lookup subtasks, and it returns structured output the orchestrator can parse.

With computer use and image input it can also drive interface-level tasks, reading screenshots to decide the next action. Structured output keeps each step machine-readable when many such calls run in sequence.

## Call it

OpenAI-compatible chat completions:

```bash
curl https://www.realrelay.ai/v1/chat/completions \
  -H "Authorization: Bearer $REALRELAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "openai/gpt-5.4-mini", "messages": [{"role": "user", "content": "Hello"}]}'
```

## FAQ

### What is GPT-5.4 mini?

GPT-5.4 mini is a compact model from OpenAI designed for coding, computer use and subagents.

### How much does GPT-5.4 mini cost?

GPT-5.4 mini costs $0.75 per 1M input tokens and $4.50 per 1M output tokens. Cached input reads cost $0.075 per 1M tokens.

### What is the context window of GPT-5.4 mini?

GPT-5.4 mini has a 400K-token context window and can generate up to 128,000 output tokens.

### What inputs and outputs does GPT-5.4 mini support?

GPT-5.4 mini accepts text and image input and returns text.

### Does GPT-5.4 mini support tool calling?

Yes. GPT-5.4 mini supports tool calling, along with reasoning, structured output and image input.

### When was GPT-5.4 mini released?

GPT-5.4 mini was released on March 17, 2026. Its knowledge cutoff is August 31, 2025.

### How do I call GPT-5.4 mini via the API?

Call GPT-5.4 mini through RealRelay's OpenAI-compatible API by setting the model name to openai/gpt-5.4-mini.