Model board / MiniMax M3

MiniMax M3

latest MiniMax

Verified 2026-09-03 against the pages listed below. Prices are USD per 1M tokens, list rate — no committed-use, enterprise or volume discount.

Input
$0.3
Cached input
$0.06
Cache write
Output
$1.20
Context window
1,000,000
Max output
262,144
Weights
unverified
Caching
explicit

host price (Together). OpenRouter still lists $0.23/$0.96 with no discount label, and now also states a $0.05 cache-read rate not previously on this board. The 512K input step recorded on 2026-07-30 was not restated on either page again this pass

1M advertised, but priced in two tiers — the step is at 512K input, not at the window. OpenRouter states 1,048,576 input and 262,144 max completion tokens; the max output is new to this row this pass

Watch out

The cheap rate is promotional and only holds below 512K input; open-weight status unverified

What we would use it for

Cheap general-purpose model with an 80% cached-input discount

What this has cost, week by week

Output price at each refresh since 2026-07-30. A vendor change is the provider changing its price. Our correction means this board changed what it prints and the provider's charge did not move — the two are never merged, because publishing the second as the first would be a price rise that never happened.

07-30 $1.2008-06 $1.2008-14 $1.2008-20 $1.2008-27 $1.2009-03 $1.20
newly stated2026-09-03 · max_output — 262,144 stated on the OpenRouter model page this row already cites. The price is unchanged: Together still lists $0.30/$0.06/$1.20, which is the rate this row prints.

Where these numbers came from

No number on this page was written that was not read from one of those pages on 2026-09-03. If one is wrong, send us the source URL and it is corrected in the next weekly refresh, with a changelog line saying what was wrong — contact@naderu.com.

The rest of the board

This row sits alongside 53 others on the model board, refreshed every week. The same data as JSON, and the terms for using it, are at /board-data.