MiniMax M3 API

MiniMax M3 is an open-weight model released by MiniMax in May 2026. It has around 428 billion parameters, with about 23 billion active per token, and uses MiniMax's own sparse attention design, which the company says makes it much faster than M2 at long context. MiniMax built it for coding, agentic work and professional tasks, and says it is natively multimodal and can operate a desktop computer. In Jantaris it has a 1,000,000-token context window and up to 512,000 output tokens.

Best for

• Budget coding agents with many tool calls
• Long outputs such as complete files, reports or datasets
• Research and browsing-style agent tasks
• Long-context processing at a low cost per token

How to call it

Use the OpenAI SDK with Jantaris as the base URL:

from openai import OpenAI

client = OpenAI(base_url="https://jantaris.com/v1", api_key="sk-jt-...")
resp = client.chat.completions.create(
    model="minimax-m3",
    messages=[{"role": "user", "content": "Write a Python CLI that converts CSV files to Parquet."}],
)
print(resp.choices[0].message.content)

FAQ

How much does MiniMax M3 cost?

$0.81 / $3.24 per 1M tokens in Jantaris. It sits in the low-price group of the catalog, a reasonable pick when an agent sends dozens of requests per task.

Can MiniMax M3 turn thinking off?

MiniMax supports toggling thinking on or off for M3. With thinking on, answers take longer and use more output tokens, so for simple rewrites or extraction a shorter prompt with clear instructions keeps costs down.

How does MiniMax M3 compare with DeepSeek V4.1 Flash?

Both are open-weight, low-cost models with a 1M-token context aimed at coding and agents. M3 allows longer single replies (512K versus 384K tokens). The fastest way to choose is to run the same agent task on M3 and DeepSeek V4.1 Flash in Jantaris and compare quality and cost.

How to call MiniMax M3 via API

from openai import OpenAI

client = OpenAI(base_url="https://jantaris.com/v1", api_key="sk-jt-...")
resp = client.chat.completions.create(
    model="minimax-m3",
    messages=[{"role": "user", "content": "Write a haiku about Baltic amber."}],
)
print(resp.choices[0].message.content)

curl https://jantaris.com/v1/chat/completions \
  -H "Authorization: Bearer $JANTARIS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "minimax-m3", "messages": [{"role": "user", "content": "Write a haiku about Baltic amber."}]}'

import OpenAI from "openai";

const client = new OpenAI({ baseURL: "https://jantaris.com/v1", apiKey: process.env.JANTARIS_API_KEY });
const resp = await client.chat.completions.create({
  model: "minimax-m3",
  messages: [{ role: "user", content: "Write a haiku about Baltic amber." }],
});
console.log(resp.choices[0].message.content);