MiniMax-M3 vs Beam 1.0: speed, latency and cost

MiniMax-M3 leads the intelligence index 18.9 to 17.7. MiniMax-M3 costs 13% less on blended price. Compare price, speed, latency and a clear verdict.

Metric	MiniMax-M3	Beam 1.0
Intelligence Index	18.9	17.7
Input price (per 1M tokens)	$0.30	$0.40
Output price (per 1M tokens)	$1.20	$1.20
Output speed (tokens/s)	121	344
Time to first token	0.82s	0.70s
Release date	2026-06-01	2026-10-16

Metric

MiniMax-M3

Beam 1.0

Intelligence Index

18.9

17.7

Input price (per 1M tokens)

$0.30

$0.40

Output price (per 1M tokens)

$1.20

Output speed (tokens/s)

121

344

Time to first token

0.82s

0.70s

Release date

2026-06-01

2026-10-16

There is very little daylight between MiniMax-M3 (MiniMax) and Beam 1.0 (Reflection) on the overall index. All numbers are the catalogue's recorded values; nobody ran either model by hand for this page. Beam 1.0 streams output at about 344 tokens per second, 2.9 times the rate of MiniMax-M3.

Beam 1.0 is the faster writer

Measured output speed is 121 tokens per second for MiniMax-M3 and 344 tokens per second for Beam 1.0. Beam 1.0 is ahead by 185% on throughput. The wait for the first token averages 0.82 seconds on MiniMax-M3 against 0.70 seconds on Beam 1.0. MiniMax-M3 waits 1.2 times as long before its first token.

Add the two numbers up for a 1,000-token reply and Beam 1.0 finishes first: 9.1 seconds for MiniMax-M3, 3.6 seconds for Beam 1.0.

How the intelligence index compares

On the intelligence index, MiniMax-M3 scores 18.9 against 17.7 for Beam 1.0. It is a small margin for an index that averages many evaluations.

Among 109 active scored models, MiniMax-M3 sits 79th and Beam 1.0 84th.

Specialist coding and math indexes are missing for both, so treat the overall score as the only capability signal here.

Release timing

The release dates are June 1, 2026 for MiniMax-M3 and October 16, 2026 for Beam 1.0, about 5 months apart.

Token pricing and a sample bill

Per million tokens, MiniMax-M3 charges $0.30 for input and $1.20 for output, while Beam 1.0 charges $0.40 and $1.20. Output rates are close, so the input rate does most of the work in favour of MiniMax-M3.

Take 1 million input tokens and 200,000 output tokens: that comes to $0.54 on MiniMax-M3 and $0.64 on Beam 1.0. Scale it up to classifying 100,000 short messages at 150 input tokens and 5 output tokens each and the bill is $5.10 for MiniMax-M3 against $6.60 for Beam 1.0. Blended at three input tokens per output token, the effective rates are $0.525 for MiniMax-M3 and $0.60 for Beam 1.0.

Is the price in proportion?

Per point of intelligence index, MiniMax-M3 costs $0.028 of blended price per million tokens and Beam 1.0 $0.034. With the higher score and the lower cost per point, MiniMax-M3 is the better value per point on these numbers.

Quick answers

Is MiniMax-M3 better than Beam 1.0?

It depends on what you need: MiniMax-M3 scores higher on the intelligence index, 18.9 to 17.7; MiniMax-M3 is 13% cheaper on blended price; and Beam 1.0 streams output 2.9 times as fast.

Are MiniMax-M3 and Beam 1.0 from the same company?

MiniMax-M3 is made by MiniMax and Beam 1.0 by Reflection. MiniMax-M3 was released on June 1, 2026 and Beam 1.0 on October 16, 2026.

Which is faster, MiniMax-M3 or Beam 1.0?

Beam 1.0 generates output faster, at 344 tokens per second against 121 tokens per second. Beam 1.0 reaches its first token sooner (0.70 seconds against 0.82 seconds).

How much do MiniMax-M3 and Beam 1.0 cost?

MiniMax-M3 is cheaper on a blended basis. MiniMax-M3 lists $0.30 per million input tokens and $1.20 per million output tokens; Beam 1.0 lists $0.40 and $1.20. For 1 million input and 200,000 output tokens that is $0.54 against $0.64.

Which is newer, MiniMax-M3 or Beam 1.0?

Beam 1.0 is newer. It was released on October 16, 2026, about 5 months after MiniMax-M3 (June 1, 2026).

Final whistle

Short answer: on balance the numbers favour Beam 1.0, though MiniMax-M3 keeps a case of its own. Pick MiniMax-M3 if you need the higher overall score (18.9 against 17.7) and cost matters ($1.20 against $1.20 per million output tokens). Pick Beam 1.0 if long outputs need to finish quickly and you want the more recent release.

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 MiniMax-M3 and Beam 1.0 pages. The AI Benchmark methodology explains how that data is collected and shown.