DeepSeek V4 Pro 0813 Max vs MiniMax-M3 compared

DeepSeek V4 Pro 0813 Max leads the intelligence index 23.1 to 18.9. MiniMax-M3 costs 73% less on blended price. See token costs, speed and which to pick.

Metric	DeepSeek V4 Pro 0813 (Max)	MiniMax-M3
Intelligence Index	23.1	18.9
Input price (per 1M tokens)	$1.32	$0.30
Output price (per 1M tokens)	$3.96	$1.20
Output speed (tokens/s)	67	121
Time to first token	1.7s	0.82s
Release date	2026-08-13	2026-06-01

Metric

DeepSeek V4 Pro 0813 (Max)

MiniMax-M3

Intelligence Index

23.1

18.9

Input price (per 1M tokens)

$1.32

$0.30

Output price (per 1M tokens)

$3.96

$1.20

Output speed (tokens/s)

121

Time to first token

1.7s

0.82s

Release date

2026-08-13

2026-06-01

DeepSeek V4 Pro 0813 (Max) (DeepSeek) and MiniMax-M3 (MiniMax) meet here on their recorded benchmark, price and speed data. Everything below is read from the benchmark, price and speed data in Franklin AI's catalogue; nothing here comes from a hands-on test. The clearest gap is on the bill: DeepSeek V4 Pro 0813 Max is 3.8 times the blended price of MiniMax-M3.

Running costs

DeepSeek V4 Pro 0813 Max lists $1.32 per million input tokens and $3.96 per million output tokens; MiniMax-M3 lists $0.30 and $1.20. That makes MiniMax-M3's output 70% cheaper.

Take 1 million input tokens and 200,000 output tokens: that comes to $2.11 on DeepSeek V4 Pro 0813 Max and $0.54 on MiniMax-M3. A bigger example, an agent that reads 50 million tokens and writes 5 million in a month, costs $85.80 on DeepSeek V4 Pro 0813 Max and $21.00 on MiniMax-M3: a gap of $64.80.

MiniMax-M3 is the faster writer

DeepSeek V4 Pro 0813 Max generates about 67 tokens per second and MiniMax-M3 about 121 tokens per second. That makes MiniMax-M3 roughly 1.8 times as fast at writing. Before any text appears, DeepSeek V4 Pro 0813 Max takes 1.7 seconds and MiniMax-M3 0.82 seconds. In an interactive app, MiniMax-M3 will feel noticeably more responsive.

Put together, a 400-token answer would take about 7.7 seconds from DeepSeek V4 Pro 0813 Max and 4.1 seconds from MiniMax-M3, counting the first-token wait.

Overall capability: 4.2 points apart

The catalogue puts DeepSeek V4 Pro 0813 Max at 23.1 and MiniMax-M3 at 18.9 on its intelligence index. A lead of this size is hard to dismiss as noise. In relative terms, DeepSeek V4 Pro 0813 Max scores 22% higher.

Across the 109 active models with an intelligence score in the catalogue, DeepSeek V4 Pro 0813 Max ranks 72nd and MiniMax-M3 ranks 79th.

For coding or maths work specifically, the catalogue has no dedicated index on either side yet.

Which is newer

DeepSeek V4 Pro 0813 Max arrived on August 13, 2026, about 2 months after MiniMax-M3 (June 1, 2026).

Is the price in proportion?

Per point of intelligence index, DeepSeek V4 Pro 0813 Max costs $0.086 of blended price per million tokens and MiniMax-M3 $0.028. DeepSeek V4 Pro 0813 Max costs 3.1 times as much per point, which is the price of its lead.

Frequently asked questions

Is MiniMax-M3 better than DeepSeek V4 Pro 0813 Max?

Not across the board: DeepSeek V4 Pro 0813 Max scores higher on the intelligence index, 23.1 to 18.9; MiniMax-M3 is 73% cheaper on blended price; and MiniMax-M3 streams output 1.8 times as fast.

How much do DeepSeek V4 Pro 0813 Max and MiniMax-M3 cost?

MiniMax-M3 is cheaper on a blended basis. DeepSeek V4 Pro 0813 Max lists $1.32 per million input tokens and $3.96 per million output tokens; MiniMax-M3 lists $0.30 and $1.20. For 1 million input and 200,000 output tokens that is $2.11 against $0.54.

When were DeepSeek V4 Pro 0813 Max and MiniMax-M3 released?

DeepSeek V4 Pro 0813 Max is newer. It was released on August 13, 2026, about 2 months after MiniMax-M3 (June 1, 2026).

Was this comparison tested by hand?

The figures are the recorded benchmark, price and speed data in Franklin AI's model catalogue. This page is written from that data automatically; neither model was tested by hand for it.

Which is faster, DeepSeek V4 Pro 0813 Max or MiniMax-M3?

MiniMax-M3 generates output faster, at 121 tokens per second against 67 tokens per second. MiniMax-M3 reaches its first token sooner (0.82 seconds against 1.7 seconds).

Final whistle

Short answer: on balance the numbers favour MiniMax-M3, though DeepSeek V4 Pro 0813 Max keeps a case of its own. Pick DeepSeek V4 Pro 0813 Max if you need the higher overall score (23.1 against 18.9) and you want the more recent release. Pick MiniMax-M3 if cost matters ($1.20 against $3.96 per million output tokens) and long outputs need to finish quickly.

Where the numbers come from

The scores, speed and prices in this comparison come from independent third-party model evaluations that Franklin AI syncs each day, and the article was written from a snapshot of them. For the latest results, open the DeepSeek V4 Pro 0813 (Max) and MiniMax-M3 pages. The AI Benchmark methodology explains how that data is collected and shown.