OpenAI's decision to slash GPT-5.6 Luna pricing by 80% — from $1.00 to $0.20 per million input tokens — represents the most aggressive price cut in the company's history. The move fundamentally alters the economics of AI deployment. [1]

The new pricing makes AI inference cost-competitive with human labor for an expanding range of tasks. At $0.20 per million tokens, processing a 10,000-word document costs approximately $0.004 — less than the cost of a single keystroke. The implications for knowledge work are profound. [1]

The price cut reflects OpenAI's confidence in its inference efficiency gains. The company has reduced per-token costs through architectural improvements, hardware optimization, and scale economies. The pricing strategy aims to capture market share from competitors who cannot match the new price points. [2]

Competitors face a difficult choice: match OpenAI's pricing and accept margin compression, or maintain higher prices and watch market share erode. For startups building on AI infrastructure, the cost reduction expands what is economically viable. Tasks that were marginally unprofitable at $1.00 per million tokens become clearly viable at $0.20. [2]

The broader AI market is experiencing a deflationary spiral. As inference costs fall, use cases expand, driving further scale economies. The question is whether OpenAI can sustain this pricing while maintaining the quality that justifies it. [1]

-- KENJI NAKAMURA, Tokyo