MiniMax H3 and H3 Max are 50% off on AI Gateway

Vercel AI Gateway is temporarily reducing costs for MiniMax video models by 50% through a limited-time discount. The promotion covers both H3 and H3 Max, which serve different use cases: H3 targets higher 2K resolution output, while H3 Max trades resolution for faster rendering at 480p and 768p. Engineers can choose between the two based on whether their priority is quality or speed without worrying about different pricing tiers during this period.

The discount requires no integration changes, as the model IDs remain unchanged. Existing requests automatically qualify for the lower rate, which simplifies adoption for teams already using these models. This approach reduces friction for engineers who want to take advantage of the promotion without refactoring their code or adjusting their workflows. The only requirement is that requests must be billed through Vercel AI Gateway, which may limit the discount’s applicability for teams using other billing methods.

The models support a range of input types, including text prompts, starting images, frame pairs, and reference media. This flexibility allows engineers to experiment with different creative workflows without additional cost during the promotion. However, the rendering process remains time-intensive, with generation taking minutes rather than seconds. The asynchronous polling mechanism helps manage this latency, but it may still be a constraint for applications requiring real-time video generation.

The discount is available for a fixed two-week window, which may influence how teams prioritize its use. Engineers working on projects with tight deadlines or budget constraints could benefit from accelerating their video generation tasks during this period. However, the temporary nature of the promotion means that long-term cost planning should still account for the standard pricing once the discount expires.