GPT-5.4 is OpenAI’s most capable model to date. It outperforms all other models on our internal benchmarks. For cost-sensitive workflows where raw intelligence matters less, GPT-5.3 Codex remains a strong option at a lower price point.

Strengths

  • GPT-5.4 is the leading model on our internal evals.
  • Confident and decisive. It works through messy, ambiguous problems without second-guessing and stays proactive about next steps.
  • Strong at parallelizing work and maintaining momentum across long workflows.

Limitations

  • Higher per-token cost than Codex 5.3. If cost per unit of work matters more than peak intelligence, Codex 5.3 may still be the better fit.
  • Output can read as more sprawling depending on taste. It favors thoroughness over brevity.
  • New model; early feedback and taste tests are still developing.

Tools

GPT-5.4 has access to all agent tools when used with Cursor including:

Learn more about how tools work and tool calling fundamentals.

Pricing

Cursor plans include two usage pools. GPT-5.4 draws from the third-party Other Models pool, which charges at the rates below. All prices are per million tokens.

A Fast mode variant (gpt-5.4-fast) is available for priority processing. It runs 15% faster at 2x the standard price. When input exceeds 272k tokens (long context), input pricing doubles and output pricing is 1.5x the standard rate.