MiniMax-M2.7

Parameters

229B

Published

Mar 18, 2026

MiniMax-M2.7 is an AI model developed by MiniMax (China), first published in March 2026. It works in the language domain, on tasks such as language modeling/generation.

Epoch AI has no training-compute estimate for this model. The model has 229,000,000,000 parameters.

Access: Open weights (non-commercial). Its weights are openly available. Epoch AI rates the confidence of this record as confident.

Full record

Organization: MiniMax
Country of organization: China
Domain: Language
Task: Language modeling/generation
Parameters: 229,000,000,000
Model accessibility: Open weights (non-commercial)
Open weights: Yes
Epoch confidence: Confident

Benchmark results

01	WebDev Arena	Arena Score	1400.5
02	ALE Bench	Performance	599.3
03	Epoch Capabilities Index	ECI Score	145.7
04	SciCode	Score	0.47
05	Terminal-Bench	Accuracy mean	0.45
06	WeirdML	Accuracy	0.37
07	Exploitbench	Mean capability	0.13
08	Proofbench	Accuracy	0.03
09	Critpt	Accuracy	0.01
10	Gbaeval	Overall score	0