MiniMax API — Access MiniMax-M2.5 via AIsa
MiniMax API: Access MiniMax-M2.5
MiniMax is one of China's most well-funded independent AI labs, and MiniMax-M2.5 is its flagship text model — a large mixture-of-experts architecture with a 196,608-token context window, strong multilingual performance, and competitive reasoning capability.
Through AIsa, you access MiniMax-M2.5 with a single OpenAI-compatible API key. No MiniMax account, no separate billing, no rate-limit management on your end.
Supported MiniMax models
| Model | Context window | Best for | Input price* | Output price* |
|---|---|---|---|---|
MiniMax-M2.5 | 196,608 tokens | Long-context reasoning, multilingual tasks, document analysis | $0.21/M | $0.84/M |
* See marketplace.aisa.one/pricing for current AIsa rates.
Quickstart
Python
from openai import OpenAI
client = OpenAI(
api_key="YOUR_AISA_API_KEY",
base_url="https://api.aisa.one/v1"
)
response = client.chat.completions.create(
model="MiniMax-M2.5",
messages=[
{"role": "user", "content": "Summarise the key arguments in this research paper."}
]
)
print(response.choices[0].message.content)Node.js
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.AISA_API_KEY,
baseURL: "https://api.aisa.one/v1",
});
const response = await client.chat.completions.create({
model: "MiniMax-M2.5",
messages: [
{ role: "user", content: "Draft a detailed competitive analysis of the EV market in Southeast Asia." }
],
});
console.log(response.choices[0].message.content);Streaming
stream = client.chat.completions.create(
model="MiniMax-M2.5",
messages=[{"role": "user", "content": "Write a thorough technical breakdown of retrieval-augmented generation."}],
stream=True
)
for chunk in stream:
if chunk.choices[0].delta.content:
print(chunk.choices[0].delta.content, end="", flush=True)Model guide
MiniMax-M2.5 — long context, strong multilingual
MiniMax-M2.5 is built around a mixture-of-experts architecture that gives it efficient inference relative to its capability level. Its 196K context window sits between DeepSeek V3.2 (128K) and Kimi K2.5 (256K), making it well-suited for tasks that require processing large documents, long conversation histories, or extended structured data — without reaching for the most expensive frontier context tiers.
Use when you need:
- Long document processing — contracts, reports, research papers — in a single pass
- Strong multilingual reasoning, especially Chinese-English bilingual tasks
- Cost-efficient processing across large batches of medium-length content
- A capable general-purpose model with broad coverage across text task types
# Long document processing
with open("annual_report.txt") as f:
document = f.read()
response = client.chat.completions.create(
model="MiniMax-M2.5",
messages=[
{"role": "system", "content": "You are a financial analyst. Extract key risks, opportunities, and financial highlights."},
{"role": "user", "content": f"Analyse this annual report:\n\n{document}"}
]
)
# Multilingual tasks
response = client.chat.completions.create(
model="MiniMax-M2.5",
messages=[
{"role": "user", "content": "请将以下英文合同条款翻译成中文,并标注任何可能存在法律风险的部分。\n\n[contract text here]"}
]
)
# Extended conversation
messages = [{"role": "system", "content": "You are a senior product strategist."}]
messages.append({"role": "user", "content": "Let's build a go-to-market plan for our new B2B SaaS product..."})
response = client.chat.completions.create(
model="MiniMax-M2.5",
messages=messages
)Function calling
MiniMax-M2.5 supports function calling with the standard OpenAI tool-calling schema:
tools = [
{
"type": "function",
"function": {
"name": "search_documents",
"description": "Search an internal document database and return relevant passages",
"parameters": {
"type": "object",
"properties": {
"query": {"type": "string", "description": "Search query"},
"max_results": {"type": "integer", "description": "Maximum number of results to return", "default": 5}
},
"required": ["query"]
}
}
}
]
response = client.chat.completions.create(
model="MiniMax-M2.5",
messages=[{"role": "user", "content": "Find all documents related to our Q3 revenue projections."}],
tools=tools,
tool_choice="auto"
)Context window: how MiniMax-M2.5 compares
| Model | Context window | Provider |
|---|---|---|
qwen3.6-plus | 1,000,000 tokens | Alibaba |
kimi-k2.5 | 256,000 tokens | Moonshot AI |
MiniMax-M2.5 | 196,608 tokens | MiniMax |
qwen3-max | 262,144 tokens | Alibaba |
deepseek-v3.2 | 128,000 tokens | DeepSeek |
seed-1-6-250915 | 131,072 tokens | ByteDance |
For most document and conversation tasks that don't require million-token context, MiniMax-M2.5's 196K window is sufficient and cost-efficient.
Data privacy
MiniMax-M2.5 is accessed through AIsa's enterprise agreement with MiniMax. Customer data is not used for model training. For compliance requirements, contact us.
What's next
- All Chinese AI models — full model comparison table
- Qwen models — Alibaba's 1M-context flagship with Key Account partner pricing
- DeepSeek V3.2 — cost-efficient general use and coding
- Kimi K2.5 — 1T parameter MoE for agentic and visual coding tasks
- ByteDance Seed & Seedream — Seed series and Seedream 4.5 image generation
- GLM-5 — Zhipu AI's flagship reasoning model
Updated 8 days ago
