Skip to main content

Documentation Index

Fetch the complete documentation index at: https://visionagents.ai/llms.txt

Use this file to discover all available pages before exploring further.

MiniMax provides powerful large language models, including the latest MiniMax-M3 agentic reasoning model and the M-series lineup, through an OpenAI-compatible Chat Completions API. Use them with Vision Agents via the OpenAI plugin by pointing ChatCompletionsLLM at MiniMax’s endpoint.
Vision Agents requires a Stream account for real-time transport. Get your MiniMax API key from the MiniMax Platform.

​
Installation

uv add "vision-agents[openai]"

​
Quick Start

from vision_agents.core import Agent, User
from vision_agents.plugins import openai, getstream, deepgram, elevenlabs

agent = Agent(
    edge=getstream.Edge(),
    agent_user=User(name="Assistant", id="agent"),
    instructions="You are a helpful assistant.",
    llm=openai.ChatCompletionsLLM(
        model="MiniMax-M3",
        base_url="https://api.minimax.io/v1",
        api_key="your_minimax_api_key",  # or set MINIMAX_API_KEY env var
    ),
    stt=deepgram.STT(),
    tts=elevenlabs.TTS(),
)
Set MINIMAX_API_KEY in your environment or pass api_key directly.When deploying to Asia, pair MiniMax with the Tencent RTC edge transport for the lowest end-to-end latency, and point at the in-region MiniMax endpoint https://api.minimaxi.com/v1 using a key from platform.minimaxi.com.
from vision_agents.core import Agent, User
from vision_agents.plugins import openai, tencent, deepgram, elevenlabs

agent = Agent(
    edge=tencent.Edge(),  # low-latency edge in mainland China and Asia
    agent_user=User(name="Assistant", id="agent"),
    instructions="You are a helpful assistant.",
    llm=openai.ChatCompletionsLLM(
        model="MiniMax-M3",
        base_url="https://api.minimaxi.com/v1",
    ),
    stt=deepgram.STT(),
    tts=elevenlabs.TTS(),
)

​
Parameters

NameTypeDefaultDescription
modelstr—Model identifier (see available models below)
api_keystrNoneAPI key (defaults to MINIMAX_API_KEY env var)
base_urlstr"https://api.minimax.io/v1"MiniMax API endpoint

​
Available Models

These models are supported through the OpenAI-compatible API:
ModelContextDescription
MiniMax-M31MLatest M-series model for agentic reasoning, tool use, coding, and long context
MiniMax-M2.7205KRecursive self-improvement; ~60 tps output
MiniMax-M2.7-highspeed205KSame as M2.7 with faster output (~100 tps)
MiniMax-M2.5205KPeak performance for complex tasks; ~60 tps
MiniMax-M2.5-highspeed205KSame as M2.5 with faster output (~100 tps)
MiniMax-M2.1205KEnhanced multi-language programming; ~60 tps
MiniMax-M2.1-highspeed205KFaster M2.1 variant (~100 tps)
MiniMax-M2205KAgentic capabilities and advanced reasoning
For MiniMax-M3, MiniMax recommends temperature=1.0 (range [0, 2]) and top_p=0.95 (default). M2.x models default to top_p=0.9. M3 also supports multimodal input (images and videos) and optional deep thinking via the thinking parameter (adaptive by default). See the MiniMax OpenAI API reference for reasoning_split, streaming usage, and other supported parameters.

​
Function Calling

MiniMax models support function calling with automatic tool invocation: 
@agent.llm.register_function(description="Get weather for a location")
async def get_weather(location: str) -> str:
    return f"The weather in {location} is sunny and 72°F"
In multi-turn tool conversations, preserve the full assistant message (including tool_calls and any reasoning content) in the conversation history so the reasoning chain stays intact. Vision Agents handles this when using registered functions on the agent. See the Function Calling guide for details.

​
Next Steps

Build a Voice Agent

Get started with voice

Build a Video Agent

Add video processing