MiniMax M3

model="MiniMax-M3"

MiniMax-M3 is MiniMax’s frontier multimodal coding and agentic model, built on the MiniMax Sparse Attention (MSA) architecture. It supports up to a 1M-token context window and accepts image and video inputs. The model is designed for code generation, agentic workflows, tool use, long-context understanding, and multi-step reasoning, showing strong performance on benchmarks such as SWE-Bench Pro, Terminal-Bench 2.1, and MCP Atlas.

Model Capabilities

Accepts (Input)

Text Generation

Generates (Output)

Text Responses

Quick Integration

Python (OpenAI SDK)

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://helyxai.space/v1/chat/completions",
    api_key=os.environ.get("HELYX_API_KEY")
)

response = client.chat.completions.create(
    model="MiniMax-M3",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "How do I use this model?"}
    ],
    max_tokens=1048576
)

print(response.choices[0].message.content)

Architecture

Context Window

1,048,576 tokens

Max Input

131,072

Max Output

1,048,576

Usage Allowances

Free Plan (Default)

25,000 tokens / day

Automatically resets every 24 hours.

Developer Plan

100,000,000 tokens / day

Market Pricing Equivalency

Input (per 1M) $0.3000

Output (per 1M) $1.2000

Your Cost: $0.00 (Via Free Quota)

Contact Support