MiniMax's API pricing for its M3 and M2.7 models has been clarified, revealing that the standard rate for M3 is $0.30 per million input tokens and $1.20 per million output tokens, with a stated permanent 50% discount from a higher list price. Cache reads are significantly cheaper at $0.06 per million tokens for M2.7, while cache writes are more expensive than standard input. The pricing structure for extended context windows beyond 512K tokens for M3 is currently in a gated tier requiring sales contact, and faster response tiers vary by model, with M3 using a priority service tier and M2.7 offering a separate high-speed variant. Notably, most resellers offer MiniMax models at nearly identical prices, suggesting they are passing the costs through with minimal markup. AI
IMPACT Provides clarity on the cost of using MiniMax models, aiding developers in project budgeting and platform selection.
RANK_REASON Pricing details for an AI model API from a specific vendor.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
