Models·MiniMax

MiniMax M2-her: A Text Model for Roleplay

minimax/minimax-m2-her

MiniMax M2-her is a text model for roleplay and character-led conversations.

$0.30Input, per million tokens
$1.20Output, per million tokens
66KContext window
2KLongest answer

66K tokens of context — about 49,152 words, on a scale that ends at 2M.

What it can do

  • Takes text
  • Gives text

Released 23 Jan 2026 · checked 3 Oct 2026.

About it

MiniMax M2-her is a text-in, text-out model from MiniMax, made for roleplay and character-driven chat. It aims to keep a steady tone and personality across multiple turns. It can take up to 65,536 tokens of context and produce up to 2,048 tokens in one answer. Input costs $0.3 per million tokens; output costs $1.2 per million tokens.

Review

In a line

From MiniMax, M2-her handles text exchanges centered on roleplay and character-led chat. It is designed to keep a consistent tone and personality over a conversation.

What it's good at

The model is aimed at immersive roleplay and expressive conversations that continue over multiple turns. It focuses on keeping a character's tone and personality steady as the chat develops.

Price in plain terms

Input costs $0.3 per million tokens, and output costs $1.2 per million tokens. Tokens are small units of text that the model reads or writes. A page of about 500 words costs roughly 0.019 cents to read. A chat turn with about 1,000 tokens in and 500 out costs about 0.09 cents; 1,000 such turns cost about $0.90.

Context and limits

It can take up to 65,536 tokens of conversation and other text at once. That is the context window: the material available to the model while it responds. One answer can contain up to 2,048 tokens. It handles text input and produces text output.

Ways to use it

M2-her accepts text and returns text, so its stated use is text conversation. The model ID is minimax/minimax-m2-her.

Verdict

Consider it if you want a model focused on roleplay or character-based chat, especially when keeping a consistent voice across turns matters. Its text-only format and 2,048-token answer limit are constraints to weigh before choosing it. The example costs are about 0.09 cents per chat turn and $0.90 per 1,000 turns; no app, provider, or access plan is specified.

Verdict

MiniMax M2-her is a fit for people seeking roleplay and character-led text conversations, with an emphasis on a steady personality across turns. Its 65,536-token context can hold up to that much conversation text, though each answer is capped at 2,048 tokens. Example costs are about 0.09 cents per chat turn and $0.90 per 1,000 turns. Choose another option if you need image input or output, since this model is text-only. No particular access route is specified.

  • Designed for roleplay and character-driven chat.
  • Aims to maintain tone and personality across multiple turns.
  • Supports up to 65,536 tokens of context.
  • Input is $0.3 and output is $1.2 per million tokens.
  • Text-only input and output.
  • Answers are capped at 2,048 tokens.
  • No app or access route is specified.
  • Pick it ifRoleplay and character-based text conversations; Multi-turn chats where a consistent voice matters
  • Skip it ifYou need image input or output; You need answers longer than 2,048 tokens

Pricing

Input is $0.3 per million tokens and output is $1.2 per million. The examples put a page at about 0.019 cents to read, one chat turn at about 0.09 cents, and 1,000 turns at about $0.90.

Questions

What is MiniMax M2-her?

It is a text-in, text-out model from MiniMax designed for roleplay, character-driven chat, and multi-turn conversations.

How much does M2-her cost?

Input costs $0.3 per million tokens and output costs $1.2 per million tokens.

What does a chat turn cost?

A turn with about 1,000 tokens in and 500 out costs about 0.09 cents. A set of 1,000 such turns costs about $0.90.

How much text can it handle at once?

Its context window is 65,536 tokens, meaning that much conversation and other text can be available while it responds.

How long can one answer be?

It can produce up to 2,048 tokens in one answer.

Can it work with images?

No image input or output is listed; the model takes text and produces text.

Where can I use it?

The model ID is minimax/minimax-m2-her, and no particular app or access route is specified.

On these boards

More from MiniMax