Minimax

This gives Claude access to MiniMax's Chinese AI models through their API. You get three chat models (M2 and M1 for reasoning, Text-01 for speed), text-to-speech with emotion controls and multiple voice options, and text-to-video generation with camera movement commands you specify in brackets. The reasoning models show their thought process in a separate field, which is handy for debugging complex queries. Video generation is async and returns a task ID to poll. The TTS stuff is more granular than most, letting you tweak pitch, bitrate, and sample rate alongside the usual speed and volume. Function calling works on the standard models if you need tool use.