OpenAI/Model/GPT/5/4

• GPT-5.4

  • Key Features

    • OpenAI's frontier model for complex professional work
    • Supports reasoning effort levels: none, low, medium, high, and xhigh
    • Supports up to 1,050,000 tokens context window
    • Max output: 128,000 tokens
    • Knowledge cutoff: 2025/08

  • Performance Highlights

    • Positioned by OpenAI as its most capable model for professional work
    • Large-context pricing note
      • For prompts above 272K input tokens, pricing increases to 2x input and 1.5x output for the full session

  • Access Details

    • Latest model alias: gpt-5.4
    • Current snapshot: gpt-5.4-2026-03-05
    • Endpoints
      • v1/chat/completions
      • v1/responses
    • Modalities
      • Input: text and images
      • Output: text
      • Audio and video: not supported
    • Text pricing per 1M tokens
      • Input: $2.50
      • Cached input: $0.25
      • Output: $15.00
    • Regional processing endpoints are charged a 10% uplift

  • Technical Specifications

    • Supports reasoning tokens
    • Features
      • Streaming
      • Function calling
      • Structured outputs
      • Distillation
    • Tools in the Responses API
      • Web search
      • File search
      • Image generation
      • Code interpreter
      • Hosted shell
      • Apply patch
      • Skills
      • Computer use
      • MCP
      • Tool search
    • Tier 1 rate limits
      • 500 RPM
      • 500,000 TPM
      • 1,500,000 batch queue limit

  • Related Tooling

    • playwright-interactive
      • New curated Codex skill from openai/skills
      • Uses a persistent js_repl Playwright session for iterative browser and Electron debugging
      • Emphasizes both functional QA and a separate visual QA pass
      • Currently requires js_repl and danger-full-access according to the skill's SKILL.md

  • References

    • OpenAI API model docs
    • OpenAI API pricing
    • Release announcement
    • Playwright Interactive skill