Model result, rank 21 of 27

MiniMax M3 Direct Plus

MiniMax direct API · extra-high reasoning · costed at MiniMax's published M3 API rates. Public result card with the model’s overall score, lane measurements, runtime and cost telemetry, and the ranking formula.

Compare all models Download public JSON

Figure 1 Lane profile against the cohort Each bar is one measured lane on the shared 0 to 100 scale. The thin mark is the cohort median; the bright mark is the best score any ranked model reached on that lane.

Full / Agenticlane rank #23 of 27

71.96

SWE MVPlane rank #18 of 27

67.77

Hard Intelligencelane rank #19 of 27

71.80

MiniMax M3 Direct Plus
Cohort median
Cohort best

Overall score70.51

Rank #21

Full / Agentic71.96

Full rank #23

SWE MVP67.77

SWE rank #18

Hard Intelligence71.80

Hard rank #19

Measured cost$0.169

100.0% reliability

Overall

All-around publication view

Score70.51

Formulamean(Full, SWE, Hard Intelligence)

BasisMiniMax direct API · extra-high reasoning · costed at MiniMax's published M3 API rates

The overall score averages the measured major lanes while keeping each source measurement visible.

Lane 01

Full / Agentic benchmark

Final71.96

Capability90.18

Agentic81.18

Pass rate85.7%

Prompts43

This lane captures instruction following, structured behavior, tool discipline, and general agentic reliability.

Lane 02

Software engineering MVP

SWE score67.77

Focused final64.23

Capability62.62

Daily driver60.93

Prompts24

This lane is closer to implementation usefulness: source handling, architecture cleanliness, and deliverable quality.

Lane 03

Hard Intelligence diagnostic

Hard score71.80

Active inquiry17.50

Online adaptation98.50

Self-repair71.19

Authority integrity100.00

Hard Intelligence measures active inquiry, online adaptation, evidence-driven self-repair, and authority/salience integrity.

Telemetry

Runtime economics

Total cost$0.169

Cost / scored item$0.0023

Seconds / timed item21.94s

Runtime coverage100.0%

Recorded tokens / item4.9k

Token coverage100.0%

Cost, time, and token basis are normalized telemetry. They explain tradeoffs; they do not overwrite the capability score yet.

Interpretation

Why the result lands here.

The result is best read as a balanced benchmark entry: one overall score plus the lane measurements that produced it. Hard Intelligence score is 71.80 and contributes to the overall score alongside Full/Agentic and SWE. Hard Intelligence is published alongside Full/Agentic and SWE for the current ranking.