An open multimodal model with a 1M-token window at small-model prices, after a 50% price cut MiniMax has made permanent. Real per-million-token rates, a workload calculator, and the budget-tier comparisons.
| MODEL | INPUT $/1M | CACHED IN | OUTPUT $/1M | BLENDED | CONTEXT | YOUR COST |
|---|---|---|---|---|---|---|
MiniMax M3 MiniMax-M3 · >512K: $0.60/$2.40 | $0.30 | $0.06 | $1.20 | $0.525 | 1.05M | $0.90 |
MiniMax M2.7 MiniMax-M2.7 | $0.30 | $0.06 | $1.20 | $0.525 | — | $0.90 |
USD per 1M tokens from MiniMax’s official pricing page as of October 16, 2026. Blended assumes a 3:1 input:output ratio. Get keys at the MiniMax console.
MiniMax M3 costs $0.30 input / $1.20 output per 1M tokens — half its $0.60 / $2.40 list price, a discount MiniMax now labels permanent — for a reasoning model that takes images and video in a 1M-token context, with open weights. That is small-model money for a genuinely mid-tier multimodal model.
M2.7, the previous generation, sits at the same price point but is text-only and does not publish a context window. Unless you are pinned to it, M3 is the obvious pick of the two.
Cache reads bill at $0.06 per 1M tokens on both models — a fifth of fresh input. One catch on M3: prompts beyond 512K tokens bill at the full $0.60 / $2.40 rate, so the discount covers the first half of the window.
MiniMax M3 and M2.7 both cost $0.30 input / $1.20 output per 1M tokens, with cache reads at $0.06. M3 adds image and video input and a 1M-token context; its prompts beyond 512K tokens bill at $0.60 / $2.40.
Not anymore — M3’s rate launched as a promotional 50% discount, and MiniMax’s pricing page now marks it as permanent. The discount applies to prompts up to 512K tokens; longer prompts bill at the $0.60 / $2.40 list rate.
At $0.53/1M blended it ties DeepSeek V4.1 Flash and undercuts Gemini Flash-Lite ($0.85), while adding video input that DeepSeek lacks. GLM-5.3-Flash ($0.24 blended) and MiMo-V2.6-Flash ($0.18) are meaningfully cheaper with a similar modality surface.
M3 ships open weights, so yes — self-hosting and third-party inference providers are both options. The rates here are the first-party API prices from platform.minimax.io.
Also see: OpenAI API pricing · Claude API pricing · Gemini API pricing · Grok API pricing · DeepSeek API pricing · Kimi API pricing · GLM API pricing · MiMo API pricing · all 68 models
ReqKey gives every key a credit balance — validate, deduct, enforce limits, and watch per-consumer MiniMax spend in real time.