MiniMax: MiniMax M3 API

minimax-m3

MiniMax M3 API combines text, image, and video input with a 1.05M advertised context and unusually large completion capacity. It targets autonomous workflows that may generate extensive plans or artifacts. ^[1]

MiniMax M3 API — LLM2014 logic 2026-08: MiniMax-M3 (thinking) ranks #24 with a 37.37 median, 43.83 best score, and 982s average time. Configurations are separate. ^[1][2]

MODALITIES

Price$0.195 / $0.78 / 1MCONTEXT1.0486M

curl https://apineed.com/v1/chat/completions \
  -H "Authorization: Bearer $API_NEED_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"minimax-m3"}'

Playground

Try MiniMax M3 in the API Need Playground.

Provider

Current API Need routing and live API pricing. ^[1]

API Need35% off

Uptime

Total Context

1.0486M

Max Output

512K

Input

$0.195/ 1M

Output

$0.78/ 1M

Cache Read

$0.039/ 1M

Cache Write

Route priced

Pricing

Pricing uses the current live API Need configuration for this model.

Official

Provider baseline

Input $0.3/ 1MOutput $1.2/ 1M

Baseline

APINEED35% off

Prepaid credits

Input $0.195/ 1MOutput $0.78/ 1M

Availability

Recent public performance data for this model.

API statusLive metrics

API Need gateway99.29% uptime

Benchmarks

Editorial benchmark data is shown when this model has a verified match. ^[2]Intelligence uses the source’s median-score ranking. Efficiency views are APINEED calculations, not official LLM2014 rankings. Test costs and reference prices are converted at ¥7 per US dollar; they are not APINEED selling prices. Missing values are N/A. These results describe the source’s tested configurations, not guaranteed APINEED endpoint performance.


1st	GPT-5.5 (xhigh)	80.23	73.89	7.9%	534s	33,219	$27.51	$29.57
2nd	GPT-5.6 Sol (xhigh)	78.37	69.42	11.42%	381s	17,769	$14.71	$29.57
3rd	Kimi-K3 (max)	75.77	67.66	10.7%	1144s	41,432	$17.40	$15.00
4th	Claude Opus 5 (xhigh)	71.24	64.74	9.12%	437s	26,705	$18.43	$24.64
5th	GLM-5.3 (max)	74.48	63.91	14.19%	1009s	57,408	$6.43	$4.00
6th	GLM-5.3-Flash (max)	69.55	60.52	12.98%	863s	38,787	$0.22	$0.20
7th	Qwen3.8-Max (xhigh)	64.68	60.05	7.16%	1297s	65,135	$9.38	$5.14
8th	DeepSeek V4 Pro 0813(max)	70.13	59.63	14.97%	1601s	73,350	$7.92	$3.86
9th	DeepSeek-V4-Flash-Vision-Exp (max)	62.86	58.10	7.57%	974s	79,420	$2.86	$1.29
10th	Gemini 3.7 Flash (high)	66.62	57.56	13.6%	131s	26,596	$2.75	$3.70
11th	Grok 4.6 (high)	67.93	56.41	16.96%	813s	33,807	$5.60	$5.91
12th	Gemini 3.1 Pro (high)	69.35	55.84	19.48%	235s	28,338	$9.39	$11.83
13th	DeepSeek V4 Flash 0731 (max)	66.34	55.23	16.75%	881s	74,657	$2.69	$1.29
14th	GLM-5.2 (max)	66.53	54.70	17.78%	891s	46,273	$5.18	$4.00
15th	Qwen3.8-Flash (xhigh)	65.44	54.49	16.73%	844s	64,942	$0.70	$0.39
16th	GPT-5.6 Luna (xhigh)	65.92	51.62	21.69%	247s	37,885	$1.25	$1.18
17th	Qwen3.8-27B (xhigh)	58.48	47.65	18.52%	2318s	73,987	$3.55	$1.71
18th	Doubao-Seed-2.1-pro (high)	55.12	45.78	16.94%	2026s	85,238	$10.23	$4.29
19th	Muse Spark 1.2 (xhigh)	57.93	45.14	22.08%	359s	47,209	$5.54	$4.19
20th	Gemini 3.7 Flash (low)	53.50	44.06	17.64%	59s	10,673	$1.10	$3.70
21st	Claude Sonnet 5 (xhigh)	47.80	39.57	17.22%	545s	45,066	$12.44	$9.86
22nd	Tencent Hy3 (high)	57.89	39.51	31.75%	1076s	53,214	$0.85	$0.57
23rd	Qwen3.7-Plus (high)	51.36	38.74	24.57%	1254s	58,610	$1.88	$1.14
24th	MiniMax-M3	43.83	37.37	14.74%	982s	57,486	$1.93	$1.20
25th	Doubao-Seed-2.0-lite 0428 (high)	34.13	26.18	23.29%	669s	27,845	$0.40	$0.51
26th	Claude Opus 5	34.94	24.58	29.65%	156s	10,261	$7.08	$24.64
27th	Gemini 3.5 Flash Lite (high)	29.51	20.44	30.74%	192s	17,842	$1.23	$2.46
28th	Ling-3.0-flash	28.37	17.05	39.9%	735s	95,538	N/A	N/A
29th	Gemma 4 31B	21.36	16.09	24.67%	756s	15,290	$0.17	$0.39
30th	openPangu-2.0-Pro	20.02	15.75	21.33%	645s	28,215	$1.64	$2.07
31st	Qwen3.7-Plus	20.84	15.63	25%	323s	8,910	$0.29	$1.14
32nd	GPT-5.5 Instant	21.73	14.09	35.16%	28s	1,712	$1.42	$29.57
33rd	Mistral Medium 3.5	17.25	13.87	19.59%	200s	28,129	$5.82	$7.39
34th	Step-3.7-Flash	26.18	13.74	47.52%	282s	47,180	$1.53	$1.16
35th	MiMo-V2.5-Pro	26.31	13.00	50.59%	550s	33,095	$0.79	$0.86
36th	Dots3-Note Preview	18.63	12.95	30.49%	210s	28,386	N/A	N/A
37th	ERNIE 5.1	16.32	11.32	30.64%	468s	24,522	$1.77	$2.57
38th	Qwen3.8-27B	16.48	10.61	35.62%	192s	8,914	$0.43	$1.71
39th	openPangu-2.0-Flash	19.24	10.16	47.19%	583s	29,052	$0.19	$0.23
40th	LongCat-2.0	17.61	9.74	44.69%	374s	14,273	$0.46	$1.14
41st	Claude Sonnet 5	16.49	9.67	41.36%	113s	6,175	$1.70	$9.86
42nd	GLM-5.2	12.42	8.05	35.19%	24s	929	$0.10	$4.00
43rd	DeepSeek V4 Flash 0731	13.57	7.93	41.56%	53s	4,621	$0.17	$1.29
44th	Gemini 3.1 Flash Lite	9.01	7.11	21.09%	11s	707	$0.03	$1.48
45th	Doubao-Seed-2.1-pro	13.03	6.17	52.65%	319s	6,948	$0.83	$4.29
46th	Mistral Medium 3.5	8.45	5.93	29.82%	18s	2,589	$0.54	$7.39

Quick Start

Call MiniMax M3 with its real backend model ID.

Get an API key

Create an API key in the product app.

export API_NEED_API_KEY=sk-apineed-v1-...

Send a request

Call /v1/chat/completions with model minimax-m3.

TypeScript SDKPythoncURLOpenAI SDK

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.API_NEED_API_KEY,
  baseURL: "https://apineed.com/v1"
});

const result = await client.chat.completions.create({
  model: "minimax-m3",
  messages: [
    { role: "user", content: "Why is the sky blue?" }
  ]
});

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

Handle the response

Validate responses and errors in your application.

curl https://apineed.com/v1/chat/completions \
  -H "Authorization: Bearer $API_NEED_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "minimax-m3",
  "stream": true,
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ]
}'

Endpoints

Endpoints currently returned by the live pricing API.

POST/v1/chat/completions

Authorization: Bearer $API_NEED_API_KEY
Content-Type: application/json
HTTP-Referer: optional - your site URL, for rankings
X-Title: optional - your site name, for rankings
Model: minimax-m3

Parameters

Supported parameters are shown when verified metadata is available. ^[1]

No verified parameter metadata is currently available.

MiniMax M3 API Q&A

Model-specific answers for teams comparing MiniMax M3 API on capability, benchmark evidence, integration, and cost.

Where does MiniMax M3 API rank in LLM2014?

MiniMax M3 — LLM2014 logic 2026-08: MiniMax-M3 (thinking) ranks #24 with a 37.37 median, 43.83 best score, and 982s average time. Configurations are separate.

Can MiniMax M3 API process video?

Yes. The catalog lists text, image, and video input with text output.

What is the MiniMax M3 APINEED price?

Input costs $0.195, output costs $0.78, and cache reads cost $0.039 per million tokens.

Is MiniMax M3 suitable for real-time chat?

For MiniMax M3 API, run direct latency tests with realistic inputs and output limits. The benchmark’s measured average is not a production response-time guarantee.

Which endpoint serves MiniMax M3 API?

Call MiniMax M3 API through https://apineed.com/v1/chat/completions with minimax-m3 as the model value. Existing OpenAI SDK clients usually need only the APINEED base URL and key.

How do I validate MiniMax M3 API before release?

Evaluate MiniMax M3 API on representative prompts, record quality, latency, and token use, then choose reasoning settings and fallbacks from those results.

How to deploy MiniMax M3

Connect the live model to your application.

Top up

Add usage credit in the product app.

Create a key

Create and securely store an API key.

Configure the endpoint

Send requests to /v1/chat/completions.

Monitor responses

Track status, latency, and errors in your application.

More live models

MiniMax M2.7minimax/minimax-m2.7 H3minimax/minimax-h3