Gemini 3.7 Flash vs GPT-5.6 Sol

The price gap is the headline and it is large: Gemini 3.7 Flash runs $0.75 per million input tokens and $3.75 per million output against GPT-5.6 Sol's $4 and $20, roughly five times cheaper on input. Two caveats matter more than the raw numbers. First, Google's price is promotional through 31 December 2026 and rises to $1.50 and $7.50 on 1 January 2027, which cuts the advantage to about 2.7x. Second, these are different tiers. Flash is Google's fast, high-volume model, and its own flagship-class Gemini 3.1 Pro is still in preview, so this is not a like-for-like frontier comparison. On capacity the two are close, 1,048,576 tokens of context against 1.05M, but GPT-5.6 Sol doubles the maximum output at 128K against 65,536. OpenAI publishes a knowledge cutoff of 16 February 2026 for Sol; Google publishes none for 3.7 Flash.