MiniMax and Inferact Partner to Advance Production Inference for Open-Source Models

On October 23, 2026, AI inference provider Inferact announced a strategic partnership with MiniMax to advance production-grade inference for open-source models, with engineering improvements to be contributed back to the vLLM project.

According to both companies, the Inferact Platform has served MiniMax M3.1 Flash in production since the model's launch day, and requests made through the MiniMax platform are now routed to Inferact. Inferact said the partnership focuses on frontier production inference for open-source models and plans to continuously return joint engineering optimizations to the vLLM community. MiniMax also thanked the Inferact team for supporting M3.1 Flash on day one and expressed anticipation for future collaboration.

vLLM is a widely used high-throughput inference framework for large language models, and the collaboration is expected to benefit the broader open-source model ecosystem.