GPT-5.4

GPT-5.4 is OpenAI’s most capable frontier model, delivering higher-quality outputs with fewer iterations across coding, reasoning, and professional workflows.

Overview

GPT-5.4 brings the coding capabilities of GPT-5.3-Codex to the flagship frontier model. It excels at:

• Coding: Production-quality code generation, polished front-end UI, repo-specific patterns, and multi-file changes with fewer retries
• Reasoning: Configurable reasoning effort from none (fast, low-latency) to xhigh (deep multi-step reasoning)
• Agentic workflows: Reduced end-to-end time across multi-step trajectories, fewer tokens and tool calls
• Document understanding: Document-heavy and spreadsheet-heavy business workflows
• Image perception: Improved multimodal tasks and image analysis
• Long context: Up to 1,050,000 token context window

Key improvements over GPT-5.2

• Coding, document understanding, tool use, and instruction following
• Image perception and multimodal tasks
• Long-running task execution and multi-step agent workflows
• Token efficiency on tool-heavy workloads
• Agentic web search and multi-source synthesis

Reasoning effort

GPT-5.4 supports five reasoning effort levels:

For higher reasoning efforts, you may need to increase max_completion_tokens to avoid empty responses where all tokens are used on reasoning.

Verbosity

Control output length with the verbosity parameter:

• low: Concise, to-the-point answers
• medium (default): Balanced responses
• high: Detailed, comprehensive explanations

Specs

Pricing

Prompts with more than 272K input tokens are priced at 2x input and 1.5x output for the full session.

Links

• OpenAI GPT-5.4 guide
• OpenAI GPT-5.4 model page