Back to AI information
MiniMax released the M2.7 model, focusing on Agent, programming and office editing

MiniMax released the M2.7 model, focusing on Agent, programming and office editing

AI information Admin 129 views

MiniMax released MiniMax-M2.7 and simultaneously opened the MiniMax Agent, API, and Token Plan subscription portals. The official defines it as the first M2 series model that is "deeply involved in its own evolution", and said that it has achieved an 88% win rate compared to M2.5, focusing on software engineering, agent collaboration and professional office scenarios.

According to the official page, M2.7 is 56.22% in SWE-Pro and 57.0% in Terminal Bench 2, and it is said that it supports Agent Teams, tool search and multiple rounds of high-fidelity Office file editing. At this stage, the relevant achievements are mainly from official release materials, and the actual effect will still be affected by the complexity of the task, toolchain configuration, subscription plan and call method; There are also differences in speed, quota, and price between versions.

FAQs

Q: What model is MiniMax-M2.7?

A: MiniMax-M2.7 is a new generation of MiniMax model for agents, programming and office tasks.

Q: What are the main features of MiniMax-M2.7?

A: MiniMax-M2.7 focuses on improving software engineering, tool calling, agent team collaboration, and Office file editing.

Q: Where can developers use MiniMax-M2.7?

A: Developers can access and experience MiniMax-M2.7 on MiniMax Agent and MiniMax API platform.

Q: Who is MiniMax Token Plan suitable for?

A: MiniMax Token Plan is suitable for developers and teams that require fixed quotas, continuous calls, and multi-modal unified usage management.

Recommended Tools

More