# Claude Opus 4.5 vs GPT-5.4

As of Oct 15, 2026, Claude Opus 4.5 leads GPT-5.4 in 3 of 8 shared categories: Reasoning, Coding › Web dev, and Long context.

Claude Opus 4.5: $10 per 1M blended tokens.

GPT-5.4: $5.625 per 1M blended tokens.

Claude Opus 4.5: closed weights.

GPT-5.4: closed weights.

- Which to pick, General LLM: As of Oct 15, 2026, GPT-5.4 leads Claude Opus 4.5 in General LLM at #30 against #41.
- Which to pick, Reasoning: As of Oct 15, 2026, Claude Opus 4.5 leads GPT-5.4 in Reasoning at #30 against #32.
- Which to pick, Math: As of Oct 15, 2026, GPT-5.4 leads Claude Opus 4.5 in Math at #9 against #16.
- Which to pick, Science: As of Oct 15, 2026, GPT-5.4 leads Claude Opus 4.5 in Science at #16 against #68.
- Which to pick, Coding › Web dev: As of Oct 15, 2026, Claude Opus 4.5 leads GPT-5.4 in Coding › Web dev at #58 against #60.
- Which to pick, Long context: As of Oct 15, 2026, Claude Opus 4.5 leads GPT-5.4 in Long context at #20 against #40.
- Which to pick, Factuality: As of Oct 15, 2026, GPT-5.4 leads Claude Opus 4.5 in Factuality at #6 against #40.
- Which to pick, Abstraction (ARC): As of Oct 15, 2026, GPT-5.4 leads Claude Opus 4.5 in Abstraction (ARC) at #10 against #49.

As of Oct 15, 2026, Claude Opus 4.5 is a former state-of-the-art holder in Reasoning (since Dec 1, 2025; sources LMArena Text (style-controlled, hard prompts); methodology v1.0).

As of Oct 15, 2026, GPT-5.4 is a former state-of-the-art holder in General LLM (since Apr 10, 2026; sources LMArena Text (style-controlled, overall); methodology v1.0).