Claude Haiku 5.5 vs GPT-5.6 Luna

Where it leads

• Lower input and output prices, $0.10 and $0.50 vs $0.20 and $1.20 per 1M tokens
• Cheaper cached input, $0.01 vs $0.02 per 1M tokens
• Batch pricing, $0.05 input and $0.25 output per 1M tokens

Best for

• High-volume classification, extraction and routing under the 100,000-token line
• Cost-sensitive fan-out steps where a frontier model is overkill

Watch out for

• Priced by prompt length, and it is a cliff rather than a taper: a request whose prompt passes 100,000 tokens is billed at $0.50 input and $2.50 output for the whole request, five times the short rate
• Cache reads and cache writes both count toward that 100,000-token measurement, so a large cached prefix can tip a request into the higher band on its own
• It is the only current Claude model whose 1M window is not priced uniformly; every other 4.6-and-later model bills a 900k request at the same rate as a 9k one

Where it leads

• Higher GPQA Diamond score, 91.6% vs 89.6%
• Higher FrontierMath Tiers 1-3 (v2) score, 82.1% vs 75.1%
• Higher SimpleQA Verified score, 41% vs 23.8%
• Larger context window 1,050,000 vs 1,000,000 tokens

Best for

• High-volume API evaluation with published low token prices
• Subagents and batch reasoning where cost per call dominates

Watch out for

• Not in the pinned LiveBench snapshot; its independent scores here come from Epoch AI's own GPQA Diamond, FrontierMath and SimpleQA Verified runs