MiniMax M2.5 Achieves 80% SWE-Bench Score at $0.15 per Million Tokens
MiniMax M2.5 model rivals giant LLMs while remaining highly cost-competitive.
Published 12sem1 sourceNotable
Lire en français
≈ 17s
The fact
At SWE-Bench, it achieves 80% accuracy at $0.15 per million tokens.
These results make autonomous coding agents economically viable for SMEs.
Click the link to read an article on the topic: