# Claude Opus 4.8 vs GPT-6 Luna

As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in 9 of 10 shared categories: General LLM, Reasoning, Science, Coding, Coding › Agentic, Long context, Vision, Factuality, and Abstraction (ARC).

Claude Opus 4.8: $10 per 1M blended tokens.

GPT-6 Luna: $0.19999999999999998 per 1M blended tokens.

Claude Opus 4.8: closed weights.

GPT-6 Luna: closed weights.

- Which to pick, General LLM: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in General LLM at #21 against #82.
- Which to pick, Reasoning: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Reasoning at #10 against #76.
- Which to pick, Science: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Science at #27 against #33.
- Which to pick, Coding: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Coding at #12 against #17.
- Which to pick, Coding › Agentic: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Coding › Agentic at #12 against #17.
- Which to pick, Coding › Web dev: As of Oct 15, 2026, GPT-6 Luna leads Claude Opus 4.8 in Coding › Web dev at #31 against #35.
- Which to pick, Long context: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Long context at #10 against #75.
- Which to pick, Vision: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Vision at #13 against #17.
- Which to pick, Factuality: As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Factuality at #24 against #45.
- Which to pick, Abstraction (ARC): As of Oct 15, 2026, Claude Opus 4.8 leads GPT-6 Luna in Abstraction (ARC) at #12 against #22.

As of Oct 15, 2026, Claude Opus 4.8 is a former state-of-the-art holder in Coding › Web dev (since Jun 5, 2026; sources LMArena WebDev; methodology v1.0).

As of Oct 15, 2026, GPT-6 Luna is state of the art on best per dollar in Coding (since Oct 14, 2026; sources Terminal-Bench 4.0; methodology v1.0).