GPT-5.6-Luna Cut 80% Across Clouds; Z Code Pro Hikes to $80/mo

Plans Z Code raised Pro to $80/mo (from $72) and Max to $168/mo (from $160), while dropping listed Lite pricing entirely (evt 32668). Replit Core rose to $25/mo (from $20/mo) and Replit Pro's yearly listing ($90/yr) was removed (evt 32666).

Prices The GPT-5.6 family saw broad, consistent cuts across nearly every host. GPT-5.6-luna input fell $1→$0.2/MTok (-80%) on models.dev, llm-prices.com, Azure, and Azure Cognitive Services (evts 32430, 32428, 32319, 32317), with OpenRouter and OpenRouter-hosted variants cut $0.5→$0.1/MTok (-80%) (evts 32418, 32412, 32176, 32174). Amazon Bedrock's luna variant fell $1→$0.22/MTok (-78%) (evt 32295). GPT-5.6-terra input dropped $2.5→$2/MTok (-20%) across models.dev, llm-prices.com, and Azure sources (evts 32429, 32427, 32318, 32316), with Bedrock's terra at $2.5→$2.2/MTok (-12%, evt 32297) and OpenRouter terra/terra-pro at $1.25→$1/MTok (-20%, evts 32414, 32410, 32180, 32178). Bedrock's gpt-5.6-sol rose $5→$5.5/MTok (+10%, evt 32296).

Elsewhere: hyper/gpt-oss-120b output cut to $0.612/MTok (-10%, evt 32454); hyper/llama-3.3-70b-instruct input cut to $0.5126 (-15.5%, evt 32451); poolside/laguna-s-2.1 input cut to $0.09 (-10%) on both OpenRouter and models.dev (evts 32438, 32435). Notable hikes: z-ai/glm-5.2 input up to $1.19/MTok (+75.8%) on OpenRouter (evt 32196); moonshotai/kimi-k2.6 input up to $0.95 (+47.1%) on both models.dev and OpenRouter (evts 32362, 32226); qwen-2.5-7b-instruct input up to $0.1 (+150%, evt 32387).