When should I choose MiniMax M3?

You want OpenAI-compatible chat completions through a single LLM.API key — especially when you specifically need MiniMax M3.

What are limitations of MiniMax M3?

Like other API models, MiniMax M3 can be wrong, incomplete, or uneven on edge cases. Validate outputs for high-stakes use. Media/OCR/STT models additionally depend on input quality.

What modalities does MiniMax M3 support?

MiniMax M3 accepts text, image, video and produces text according to its architecture metadata on LLM.API.

Which providers serve MiniMax M3?

LLM.API currently lists: canopywave, minimax, nebius, novita. Availability can vary by region and account.

How do I call MiniMax M3 via API?

Send OpenAI-compatible requests to https://api.llmapi.ai/v1 with model "minimax-m3" and your LLM.API key. See the code snippet on this page.

Where is the canonical page for MiniMax M3?

https://llmapi.ai/models/minimax-m3/

Is MiniMax M3 a chat model?

Yes—MiniMax M3 is exposed as a chat/completions-style endpoint on LLM.API.

Does MiniMax M3 support streaming?

Yes—at least one listed provider advertises streaming.

What is the context length for MiniMax M3?

Reported context for MiniMax M3 is 1M tokens. Always verify the active provider row if multiple providers are listed.

Which endpoint do I call for MiniMax M3?

POST https://api.llmapi.ai/v1/chat/completions with model "minimax-m3" and your LLM.API key. The request and response bodies are OpenAI-compatible.

How do I switch from OpenAI to MiniMax M3 on LLM.API?

Change two lines: set base_url to https://api.llmapi.ai/v1 and api_key to your LLM.API key. The rest of your OpenAI SDK code stays the same.

Does MiniMax M3 support function calling and structured outputs?

Function calling: supported via tools. Structured outputs: supported via response_format. Reasoning effort accepts low, medium, high (default low).

MiniMax M3: Self-Hosting & Deployment Guide

Up to 5%

Call MiniMax M3 through LLM.API when you want MiniMax-family text generation with unified auth, provider choice, and production-friendly defaults.

Start Using API

What is MiniMax M3?

On LLM.API, MiniMax M3 (minimax-m3) serves as a MiniMax conversational model for products that need 1M tokens context windows and predictable token pricing. minimax-m3 provided by minimax, novita. It is wired for API access through LLM.API with OpenAI-compatible patterns.

How to run MiniMax M3 yourself

Open weights (other), 427B parameters. Below: how much memory the weights need at each quantization, which hardware fits them, and the commands to serve it.

MiniMax M3 VRAM requirements

Precision / quantization	Weights	Figure	Source
BF16 (full precision)	854.1 GB	Estimate	427B params × 2 bytes
FP8	430.7 GB	File size	EmbeddedLLM/MiniMax-M3-FP8-dynamic
AWQ 4-bit	240.9 GB	File size	spectator2026/MiniMax-M3-AWQ-int4
MLX 4-bit	241.5 GB	File size	mlx-community/MiniMax-M3-4bit
GGUF Q8_0	452.7 GB	File size	unsloth/MiniMax-M3-GGUF
GGUF Q4_K_M	264 GB	File size	unsloth/MiniMax-M3-GGUF

File size = actual download size on Hugging Face. Estimate = parameter count × bytes per weight. Weights only — the KV cache for long contexts needs extra memory; the fit tables below add 15% headroom as a rule of thumb.

GPU requirements: deploy MiniMax M3 on your hardware

Datacenter GPUsConsumer GPUsApple SiliconCloud instances

Hardware	BF16 (full precision)	FP8	AWQ 4-bit	MLX 4-bit	GGUF Q8_0	GGUF Q4_K_M
NVIDIA H200141 GB	7×	4×	2×	2×	4×	3×
NVIDIA H10080 GB	—	7×	4×	4×	7×	4×
NVIDIA A100 80GB80 GB	—	7×	4×	4×	7×	4×
NVIDIA L40S48 GB	—	—	6×	6×	—	7×

GPUs needed per variant.

Hardware	GGUF Q8_0	GGUF Q4_K_M
RTX 509032 GB	—	—
RTX 409024 GB	—	—
RTX 309024 GB	—	—

GGUF via llama.cpp; more cards or CPU offload for larger files.

Hardware	GGUF Q8_0	GGUF Q4_K_M	MLX 4-bit
Mac, 512 GBM3 Ultra	—	Fits	Fits
Mac, 192 GBM2 Ultra	—	—	—
Mac, 128 GBM4 Max / M3 Max	—	—	—
Mac, 64 GBM4 Pro / Max	—	—	—

Assumes ~75% of unified memory is usable by the GPU.

Hardware	BF16 (full precision)	FP8	AWQ 4-bit	MLX 4-bit	GGUF Q8_0	GGUF Q4_K_M
AWS p5.48xlarge8× H100 · 640 GB	—	Fits	Fits	Fits	Fits	Fits
AWS p4de.24xlarge8× A100 80GB · 640 GB	—	Fits	Fits	Fits	Fits	Fits
AWS g6e.12xlarge4× L40S · 192 GB	—	—	—	—	—	—
Google Cloud a3-highgpu-8g8× H100 · 640 GB	—	Fits	Fits	Fits	Fits	Fits
Azure ND H100 v58× H100 · 640 GB	—	Fits	Fits	Fits	Fits	Fits

Whole-instance GPU memory; public instance specs.

Serve it yourself

vLLMSGLangllama.cpp (GGUF)MLX (Mac)

pip install -U vllm
vllm serve MiniMaxAI/MiniMax-M3

pip install 'sglang[all]'
python -m sglang.launch_server --model-path MiniMaxAI/MiniMax-M3

llama-server -hf unsloth/MiniMax-M3-GGUF:Q4_K_M \
  --ctx-size 32768 --n-gpu-layers 99

pip install -U mlx-lm
mlx_lm.generate --model mlx-community/MiniMax-M3-4bit --prompt "Hello"

Commands use the official repositories above; vLLM and SGLang expose an OpenAI-compatible endpoint.

Skip the deploy — use LLM.API as your endpoint

No GPUs, no quantization trade-offs. Same model, OpenAI-compatible, up to 5% below list price.

Get API key See API code ↓

Providers

LLM.API routes MiniMax M3 to the providers below, with discounted effective rates versus list price.

List price by provider

InputOutput

minimax$0.6 in

$2.4 out

novita$0.3 in

$1.2 out

canopywave$0.355 in

$1.78 out

nebius$0.3 in

$1.2 out

Provider list prices; the LLM.API discount applies on top.

Provider	Pricing	Context	Capabilities
minimax	in $0.6; out $2.4 per 1M tokens	1M tokens	vision, tools, streaming, reasoning, web search, JSON, structured
novita	in $0.3; out $1.2 per 1M tokens	1M tokens	vision, tools, streaming, reasoning, web search, JSON
canopywave5% off	in $0.355; out $1.78 per 1M tokens	512K tokens	vision, tools, streaming, reasoning, web search
nebius	in $0.3; out $1.2 per 1M tokens	1M tokens	tools, streaming, reasoning, web search

Prices and availability from the LLMAPI catalogue, updated nightly. Last updated 11 Oct 2026.

Try this model

Test MiniMax M3 right here — free to start.

MiniMax M3

Hi! Want to test the model?

Suggestions for your first prompt

Code snippet

Call MiniMax M3 through the OpenAI-compatible API — POST /v1/chat/completions.

Switching from OpenAI? Change 2 lines.

- base_url="https://api.openai.com/v1"
- api_key="YOUR_OPENAI_KEY"
+ base_url="https://api.llmapi.ai/v1"
+ api_key="YOUR_LLMAPI_KEY"

Everything else stays the same — SDK, message format, tools, streaming. Set model to minimax-m3 and the call runs unchanged.

bash

curl https://api.llmapi.ai/v1/chat/completions \
  -H "Authorization: Bearer $LLMAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "minimax-m3",
    "messages": [
      {"role": "user", "content": "Give me three crisp launch checklist items."}
    ],
  "reasoning_effort": "low"
  }'

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_LLMAPI_KEY",
    base_url="https://api.llmapi.ai/v1",
)

resp = client.chat.completions.create(
    model="minimax-m3",
    messages=[
        {"role": "system", "content": "You are a precise product assistant."},
        {"role": "user", "content": "Give me three crisp launch checklist items."},
    ],
    reasoning_effort="low",
)
print(resp.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.LLMAPI_KEY,
  baseURL: "https://api.llmapi.ai/v1",
});

const resp = await client.chat.completions.create({
  model: "minimax-m3",
  messages: [
    { role: "user", content: "Give me three crisp launch checklist items." },
  ],
});

console.log(resp.choices[0].message.content);

stream = client.chat.completions.create(
    model="minimax-m3",
    messages=[{"role": "user", "content": "Write a haiku about latency."}],
    stream=True,
)

for chunk in stream:
    delta = chunk.choices[0].delta.content
    if delta:
        print(delta, end="", flush=True)

{
  "model": "minimax-m3",
  "messages": [
    {"role": "system", "content": "You are a precise product assistant."},
    {"role": "user", "content": "Give me three crisp launch checklist items."}
  ],
  "reasoning_effort": "low"
}

ENDPOINTS

Supported endpoints & features

Base URL https://api.llmapi.ai/v1 · model ID minimax-m3 · OpenAI-compatible.

POST /v1/chat/completionsChat Completions — supported

Feature	Status	How to set it
Function calling (tools)	Supported	`tools, tool_choice`
Parallel tool calls	Not supported	`parallel_tool_calls`
JSON mode	Supported	`response_format: {"type": "json_object"}`
Structured outputs	Supported	`response_format: {"type": "json_schema", …}`
Streaming	Supported	`stream: true`
Vision (image input)	Supported	`image_url content parts`
Web search	Supported	`web_search`
Reasoning effort	low · medium · high	`reasoning_effort: "low"`

ERRORS

Errors & fallback routing

Code	What it means	What to do
`401`	Missing or invalid API key.	Send `Authorization: Bearer YOUR_LLMAPI_KEY` and check the key is active.
`429`	Rate limited, or the upstream provider is throttling the request.	Back off and retry with jitter; the gateway also retries the request on another provider where one is available.
`5xx`	Upstream provider error or timeout.	Retry; fallback routing sends the retry to the next healthy provider for this model.

When a provider fails or throttles, the request is routed to the next provider serving this model — the ones listed in the providers table above.

MiniMax M3 in production

Best workloadsLLM.API features

What the model is good at.

Conversational UX
Maintains coherent multi-turn assistant behavior for product chat surfaces. Grounded in the model's chat role rather than generic chat claims.
Code assistance
Helps write, explain, refactor, and debug application code across common languages. Reflects MiniMax positioning for this endpoint.
Structured outputs
Can produce JSON-friendly or schema-oriented responses when prompted carefully. Relevant for `minimax-m3` workloads on LLM.API.
Long-context synthesis
Summarizes and cross-references information across large prompts when context allows. Relevant for `minimax-m3` workloads on LLM.API.
Multilingual drafting
Drafts and translates professional content across major business languages. Tuned to how teams typically call MiniMax M3.

What you get when you call MiniMax M3 through LLM.API.

Unified AI Routing
Practical: Reach MiniMax M3 and sibling models through one OpenAI-compatible endpoint.
One endpoint, every model.
Cost Control
Practical: Compare provider price points and keep spend visible as you scale MiniMax M3.
More performance, less spend.
Reliability Layer
Production: Retry and route across configured providers when a single upstream blips.
Never go dark on users.
Observability
Production: Trace prompts, tokens, and errors for MiniMax M3 alongside the rest of your stack.
See every token, everywhere.
Drop-in SDKs
Keep using familiar OpenAI client patterns with base URL https://api.llmapi.ai/v1.
Think tasks, not models.
Model Breadth
Practical: Swap MiniMax M3 for chat, media, or embedding alternatives without rewriting auth.
Scale from 1 to millions.

When to Use — When NOT to Use

Use it if...

Your prompts benefit from the model's family strengths (reasoning, speed, or cost) (MiniMax M3)
You want OpenAI-compatible chat completions through a single LLM.API key (MiniMax M3)
You need provider failover options exposed for this model id (MiniMax M3)
You need a general-purpose text model for assistants, agents, or content workflows (MiniMax M3)

Avoid if...

You need guaranteed real-time hard latency SLAs without benchmarking the provider
You require on-prem only deployment with no cloud inference
You need pure embedding, OCR, or media generation instead of chat
Your use case depends on unpublished proprietary benchmarks not listed here

LLM API rating for MiniMax M3

★ 3.6/ 5.0LLM API rating
50% production adoption · 30% category benchmark · 20% tooling & value✓ Verified benchmark · LMArena + Artificial Analysis

3.5Production adoptionActive open-weight community usage

3.4Benchmark qualityArtificial Analysis Intelligence Index 29.2 · LMArena text rank #86, Elo 1433

4.2Developer ecosystemContext window, tool and JSON support, price against the category median

Reasoning & chat 3.6Coding 3.6Vision & multimodal 3.7Tool & function calling 4.1Long context 4.1

How is this rating calculated?

Every model gets a separate 1.0–5.0 score for each category it can be used in, so a model can rate high for coding and lower for vision. The score combines three parts:

Production adoption — 50%. How widely the model is actually used in production: provider traffic, LiteLLM and gateway usage, open-weight downloads and its LMArena rank. Proven, widely used models score highest.
Category benchmark — 30%. The public benchmark that matters for that category — the Artificial Analysis Intelligence Index and LMArena Elo for reasoning, LMArena coding Elo for coding, image and document support plus vision Elo for vision, tool and JSON support for function calling, measured speed for voice.
Tooling & value — 20%. Context window, support for tools, structured outputs and streaming, and price compared with the category median.

Scores of 4.0 and above are shown in green, 3.0–3.9 in yellow, below 3.0 in red. When a model has no public data for a part, it gets a neutral value instead of an invented number.

A “Verified benchmark” mark (✓) means an independent lab — LMArena or Artificial Analysis — has actually tested that model, and the source figure is shown. Models without that mark are labelled “Estimated from published specs”: the score comes from the published context window, tool and schema support, price and adoption only, because no independent lab has tested them yet.

One exception: a brand-new top-tier model that scores very high on the independent benchmarks but has no usage history yet is rated on the benchmark (60%) and tooling & value (40%) alone, so a missing adoption figure can't hold it back. All inputs come from public sources and are refreshed weekly.

Adoption and benchmark inputs come from public sources — LMArena, Artificial Analysis Intelligence Index and published model specs. Updated weekly.

BENCHMARKS

MiniMax M3 benchmark scores

Intelligence index

This model30.

Tracked median18.

Scale: 0-100 index points

Output speed

This model92 t/s

Scale: 0-400 tokens per second

Reference price per 1M tokens

Input$0.30

Output$1.20

Bars compare input and output list prices for this model.

Intelligence index+67% vs median

MiniMax M3: Self-Hosting & Deployment Guide30

Median tracked model18

Higher is better. The median covers every model tracked by Artificial Analysis.

Output speed

92tok/s

Measured throughput; the full dial is 300 tokens per second.

Reference price / 1M tokens

Input$0.30

Output$1.20

Provider list prices, before the LLM.API discount.

Independent scores published by Artificial Analysis (source). Reference prices are provider list prices, not LLM.API pricing.

COMMUNITY

What developers say about MiniMax models

LMArena#86 · Elo 1433Arena coding#62 · Elo 1470Speed92 tokens/sSources: LMArena, Artificial Analysis

Summarised from publicly published developer and community reviews of this model family. Opinions are the sources’, not LLM.API’s, and may not be specific to MiniMax M3.

MiniMax M2 debuted as one of the highest-scoring open-weight models for coding and agentic tasks and is repeatedly recommended as a cheap Claude alternative.
Large-sample community tests (1,000+ prompts) report reliable tool calling and fast, shippable code.
Reviewers note it has since been overtaken by larger open-weight releases on general intelligence indices.

SOURCES

Frequently Asked Questions

When should I choose MiniMax M3?
You want OpenAI-compatible chat completions through a single LLM.API key — especially when you specifically need MiniMax M3.
What are limitations of MiniMax M3?
Like other API models, MiniMax M3 can be wrong, incomplete, or uneven on edge cases. Validate outputs for high-stakes use. Media/OCR/STT models additionally depend on input quality.
What modalities does MiniMax M3 support?
MiniMax M3 accepts text, image, video and produces text according to its architecture metadata on LLM.API.
Which providers serve MiniMax M3?
LLM.API currently lists: canopywave, minimax, nebius, novita. Availability can vary by region and account.
How do I call MiniMax M3 via API?
Send OpenAI-compatible requests to https://api.llmapi.ai/v1 with model "minimax-m3" and your LLM.API key. See the code snippet on this page.
What is MiniMax M3?
minimax-m3 provided by minimax, novita. It is wired for API access through LLM.API with OpenAI-compatible patterns. On LLM.API it is addressed as `minimax-m3`.
Where is the canonical page for MiniMax M3?
https://llmapi.ai/models/minimax-m3/
Is MiniMax M3 a chat model?
Yes—MiniMax M3 is exposed as a chat/completions-style endpoint on LLM.API.
Does MiniMax M3 support streaming?
Yes—at least one listed provider advertises streaming.
What is the context length for MiniMax M3?
Reported context for MiniMax M3 is 1M tokens. Always verify the active provider row if multiple providers are listed.

COMPARE

Competitive Models

Get one key to every model

Swap your API key. Keep your code.

Start for free Talk to sales

LLM API sells to AI agents - live prices and self-serve payment

MiniMax M3: Self-Hosting & Deployment Guide

What is MiniMax M3?

How to run MiniMax M3 yourself

MiniMax M3 VRAM requirements

GPU requirements: deploy MiniMax M3 on your hardware

Serve it yourself

Providers

Try this model

Code snippet

Supported endpoints & features

Errors & fallback routing

MiniMax M3 in production

Conversational UX

Code assistance

Structured outputs

Long-context synthesis

Multilingual drafting

Unified AI Routing

Cost Control

Reliability Layer

Observability

Drop-in SDKs

Model Breadth

When to Use — When NOT to Use

Use it if...

Avoid if...

LLM API rating for MiniMax M3

MiniMax M3 benchmark scores

Intelligence index

Output speed

Reference price per 1M tokens

What developers say about MiniMax models

Frequently Asked Questions

Competitive Models

DeepSeek V4.1 Flash

Qwen3 Coder Flash

Gemini 2.5 Flash

Get one key to every model