{
  "components": {
    "condition_encoder": {
      "condition_hidden_dim": 4096,
      "input_hop_length": 960,
      "input_sampling_rate": 24000,
      "num_condition_layers": 8,
      "out_dim": 2048,
      "output_hop_length": 512,
      "output_sampling_rate": 44100
    },
    "language_model": {
      "head_dim": 128,
      "hidden_size": 4096,
      "intermediate_size": 12288,
      "max_position_embeddings": 10240,
      "model_type": "qwen3",
      "num_attention_heads": 32,
      "num_hidden_layers": 36,
      "num_key_value_heads": 8,
      "rms_norm_eps": 1e-06,
      "rope_theta": 1000000,
      "tie_word_embeddings": false,
      "vocab_size": 200000
    },
    "rvq_depth_decoder": {
      "audio_vocab_size": 1024,
      "hidden_size": 4096,
      "intermediate_size": 6144,
      "max_position_embeddings": 16,
      "num_attention_heads": 16,
      "num_codebooks": 8,
      "num_layers": 4,
      "rms_norm_eps": 1e-06
    },
    "transformer": {
      "attention_head_dim": 64,
      "condition_dim": 2048,
      "ff_inner_dim": 8192,
      "fourier_embedding_dim": 256,
      "in_channels": 128,
      "num_attention_heads": 32,
      "num_layers": 36,
      "rotary_dim": 32
    },
    "vocoder": {
      "decoder_hidden_dim": 1536,
      "decoder_input_dim": 1024,
      "latent_channels": 128,
      "sampling_rate": 44100,
      "upsampling_ratios": [
        8,
        8,
        4,
        2
      ]
    }
  },
  "format": "mlx-minimax-music3-v1",
  "model": {
    "audio_cfg_token_id": 151654,
    "audio_code_offset": 151675,
    "audio_end_token_id": 151670,
    "audio_vocab_size": 1024,
    "frame_rate": 25,
    "latent_channels": 128,
    "latent_hop_length": 512,
    "max_audio_frames": 9000,
    "max_prompt_tokens": 5000,
    "num_codebooks": 8,
    "sampling_rate": 44100,
    "semantic_vocab_size": 16384
  },
  "quantization": {
    "bits": 8,
    "group_size": 64,
    "mode": "affine"
  }
}