MiniMax M2.5发布：性能比肩Claude Opus 4.6，输入价格约0.3美元/百万Token

Core Insights - MiniMax has launched its new text model, MiniMax M2.5, which shows significant improvements in programming capabilities and performance metrics compared to its predecessor [1][4] - The model has achieved a score of 80.2% on the SWE-Bench Verified leaderboard and 51.3% on the Multi-SWE-Bench, surpassing Opus 4.6 in multi-language complex environments [1][4] - M2.5 demonstrates "native Spec capability," allowing it to proactively decompose architecture and functional planning before coding, mimicking the work patterns of real architects [1][4] Performance and Cost Efficiency - The M2.5-lightning version supports over 100 transactions per second (TPS) output speed, approximately double that of mainstream models [2][5] - Input costs are around $0.3 per million tokens, while output costs are about $2.4 per million tokens, making it cost-effective for continuous operation [2][5] - The theoretical cost for running four agents continuously for a year is approximately $10,000, indicating a potential shift in the economic model for large-scale agent deployment [6] User Adoption and Deployment - M2.5 has been integrated into MiniMax Agent and has been globally open-sourced for localized deployment [6] - Within a day of launch, users worldwide have created over 10,000 experts on the MiniMax Agent platform, with rapid growth continuing [6]