MiniMax AI Boosts AI Innovation

Imagine a world where you have the power of high-performance AI models, but at a fraction of the cost. This is the intriguing possibility presented by MiniMax M2.5, a groundbreaking coding AI model discussed in a recent video by Better Stack. Released on February 13, 2026, MiniMax M2.5 offers performance similar to Claude Opus 4.6—with its 230 billion parameters mixture-of-experts model that activates just 10 billion crucial ones per run—yet costs about 10% of Opus’s price. It’s priced at a modest $0.15 per million input tokens and $1.20 for output tokens. The allure lies in its open weights and MIT license, which eliminate vendor lock-in and allow local deployment. However, achieving performance comes at the expense of time, as MiniMax takes longer to complete tasks compared to its counterpart Opus 4.6.

During demonstrations, while Opus swiftly executed tasks with visually refined results, MiniMax slightly lagged, producing outputs lacking certain UI elements such as dynamic task labels. Yet, it made up for this with full functionality and much lower costs. Arguably, more significant are its reinforcement learning capabilities, leading to 20% fewer tool calls and reduced token waste. In practical terms, the model supports multifile edits and interfaces seamlessly with local clusters or CI pipelines, hinting at massive workflow efficiencies for developers.

The sweetness of this disparity in cost deepens when considering MiniMax’s support for O Lama and GitHub integrations, essentially democratizing access to advanced agent functionalities. Developers can fine-tune it sans lock-in, quite a feat for a model providing nearly the same benchmark scores as Opus. These benchmarks validate MiniMax’s close competition, with only a slight gap in performance, primarily on speed.

Looking ahead, the decision to opt for MiniMax over Opus hinges on weighing the benefits of comparable agent performance against pricing and flexibility. For many, the economic case is compelling enough to switch—especially given its expedient response to market needs, with promises of even greater innovations ahead.