MiMo API

This is the current flagship pay-as-you-go text model in the official docs. It is priced domestically at ¥0.025 for cache hits, ¥3 for uncached input, and ¥6 for output per 1M tokens, or $0.0036, $0.435, and $0.87 overseas, which makes it better suited to complex engineering, critical content generation, and high-value flows where per-call quality matters more.

This is the current mainline standard pay-as-you-go text model, priced domestically at ¥0.02 for cache hits, ¥1 for uncached input, and ¥2 for output per 1M tokens, or $0.0028, $0.14, and $0.28 overseas. Compared with the flagship tier, it is better suited to day-to-day large-scale text usage, knowledge assistants, and routine coding support.

The docs list this as the current ASR model, and unlike text models it is billed by input audio duration at ¥0.5/hour domestically or $0.074/hour overseas. For meeting notes, voice input, and support-call analysis, it acts as the key speech-ingestion capability inside the MiMo platform.

VoiceClone is one of the explicitly listed TTS-series models in the official docs and is currently temporarily free. It is more suitable for turning text results into speech that feels more personalized, character-driven, or closer to a specific vocal identity, which makes it valuable for voice assistants, companion-style interactions, and dubbing workflows.

VoiceDesign is also a currently supported and temporarily free TTS-series model. It is better suited to scenarios that need speech-style design, narrated product experiences, or voice-product prototypes, showing that MiMo’s pay-as-you-go platform covers not only text and ASR but also the speech-output layer.