xinference 1.5.0.post2__py3-none-any.whl → 1.5.1__py3-none-any.whl

xinference/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2025-04-21T17:53:44+0800",
+ "date": "2025-04-30T21:28:49+0800",
  "dirty": false,
  "error": null,
- "full-revisionid": "a5d4be9f970137bde1d402420f71961826392224",
- "version": "1.5.0.post2"
+ "full-revisionid": "1c11c609971e5a5095ce8be73f0e1bba04a3132f",
+ "version": "1.5.1"
 }
 '''  # END VERSION_JSON
 

xinference/api/restful_api.py

@@ -56,6 +56,7 @@ from ..constants import (
     XINFERENCE_DEFAULT_CANCEL_BLOCK_DURATION,
     XINFERENCE_DEFAULT_ENDPOINT_PORT,
     XINFERENCE_DISABLE_METRICS,
+    XINFERENCE_SSE_PING_ATTEMPTS_SECONDS,
 )
 from ..core.event import Event, EventCollectorActor, EventType
 from ..core.supervisor import SupervisorActor
@@ -1338,7 +1339,9 @@ class RESTfulAPI(CancelMixin):
                 finally:
                     await model.decrease_serve_count()
 
-            return EventSourceResponse(stream_results())
+            return EventSourceResponse(
+                stream_results(), ping=XINFERENCE_SSE_PING_ATTEMPTS_SECONDS
+            )
         else:
             try:
                 data = await model.generate(body.prompt, kwargs, raw_params=raw_kwargs)
@@ -1606,7 +1609,9 @@ class RESTfulAPI(CancelMixin):
                         await model.decrease_serve_count()
 
                 return EventSourceResponse(
-                    media_type="application/octet-stream", content=stream_results()
+                    media_type="application/octet-stream",
+                    content=stream_results(),
+                    ping=XINFERENCE_SSE_PING_ATTEMPTS_SECONDS,
                 )
             else:
                 return Response(media_type="application/octet-stream", content=out)
@@ -2122,7 +2127,9 @@ class RESTfulAPI(CancelMixin):
                 finally:
                     await model.decrease_serve_count()
 
-            return EventSourceResponse(stream_results())
+            return EventSourceResponse(
+                stream_results(), ping=XINFERENCE_SSE_PING_ATTEMPTS_SECONDS
+            )
         else:
             try:
                 data = await model.chat(

xinference/constants.py

@@ -29,7 +29,8 @@ XINFERENCE_ENV_DISABLE_HEALTH_CHECK = "XINFERENCE_DISABLE_HEALTH_CHECK"
 XINFERENCE_ENV_DISABLE_METRICS = "XINFERENCE_DISABLE_METRICS"
 XINFERENCE_ENV_DOWNLOAD_MAX_ATTEMPTS = "XINFERENCE_DOWNLOAD_MAX_ATTEMPTS"
 XINFERENCE_ENV_TEXT_TO_IMAGE_BATCHING_SIZE = "XINFERENCE_TEXT_TO_IMAGE_BATCHING_SIZE"
-XINFERENCE_ENV_VIRTUAL_ENV = "XINFERENCE_EANBLE_VIRTUAL_ENV"
+XINFERENCE_ENV_VIRTUAL_ENV = "XINFERENCE_ENABLE_VIRTUAL_ENV"
+XINFERENCE_ENV_SSE_PING_ATTEMPTS_SECONDS = "XINFERENCE_SSE_PING_ATTEMPTS_SECONDS"
 
 
 def get_xinference_home() -> str:
@@ -89,6 +90,9 @@ XINFERENCE_DOWNLOAD_MAX_ATTEMPTS = int(
 XINFERENCE_TEXT_TO_IMAGE_BATCHING_SIZE = os.environ.get(
     XINFERENCE_ENV_TEXT_TO_IMAGE_BATCHING_SIZE, None
 )
+XINFERENCE_SSE_PING_ATTEMPTS_SECONDS = int(
+    os.environ.get(XINFERENCE_ENV_SSE_PING_ATTEMPTS_SECONDS, 600)
+)
 XINFERENCE_LAUNCH_MODEL_RETRY = 3
 XINFERENCE_DEFAULT_CANCEL_BLOCK_DURATION = 30
 XINFERENCE_ENABLE_VIRTUAL_ENV = bool(int(os.getenv(XINFERENCE_ENV_VIRTUAL_ENV, "0")))

xinference/core/supervisor.py

@@ -1102,8 +1102,8 @@ class SupervisorActor(xo.StatelessActor):
                 xavier_config=xavier_config,
                 **kwargs,
             )
-            await worker_ref.wait_for_load(_replica_model_uid)
             self._replica_model_uid_to_worker[_replica_model_uid] = worker_ref
+            await worker_ref.wait_for_load(_replica_model_uid)
             return subpool_address
 
         async def _launch_model():

xinference/core/utils.py

@@ -263,7 +263,7 @@ class CancelMixin:
     _CANCEL_TASK_NAME = "abort_block"
 
     def __init__(self):
-        self._running_tasks: weakref.WeakValueDictionary[
+        self._running_tasks: weakref.WeakValueDictionary[  # type: ignore
             str, asyncio.Task
         ] = weakref.WeakValueDictionary()
 

xinference/core/worker.py

@@ -787,9 +787,9 @@ class WorkerActor(xo.StatelessActor):
         elif model_type == "image":
             return model.model_ability
         elif model_type == "audio":
-            return [model.model_ability]
+            return model.model_ability
         elif model_type == "video":
-            return ["text_to_video"]
+            return model.model_ability
         elif model_type == "flexible":
             return ["flexible"]
         else:

xinference/deploy/cmdline.py

@@ -805,6 +805,14 @@ def remove_cache(
     type=(str, str),
     multiple=True,
 )
+@click.option(
+    "--quantization-config",
+    "-qc",
+    "quantization_config",
+    type=(str, str),
+    multiple=True,
+    help="bnb quantization config for `transformers` engine.",
+)
 @click.option(
     "--worker-ip",
     default=None,
@@ -853,6 +861,7 @@ def model_launch(
     trust_remote_code: bool,
     api_key: Optional[str],
     model_path: Optional[str],
+    quantization_config: Optional[Tuple],
 ):
     kwargs = {}
     for i in range(0, len(ctx.args), 2):
@@ -884,6 +893,12 @@ def model_launch(
     else:
         _n_gpu = int(n_gpu)
 
+    bnb_quantization_config = (
+        {k: handle_click_args_type(v) for k, v in dict(quantization_config).items()}
+        if quantization_config
+        else None
+    )
+
     image_lora_load_params = (
         {k: handle_click_args_type(v) for k, v in dict(image_lora_load_kwargs).items()}
         if image_lora_load_kwargs
@@ -929,6 +944,8 @@ def model_launch(
 
     # do not wait for launching.
     kwargs["wait_ready"] = False
+    if bnb_quantization_config:
+        kwargs["quantization_config"] = {**bnb_quantization_config}
 
     model_uid = client.launch_model(
         model_name=model_name,

xinference/model/audio/core.py

@@ -52,7 +52,7 @@ class AudioModelFamilyV1(CacheableModelSpec):
     model_revision: Optional[str]
     multilingual: bool
     language: Optional[str]
-    model_ability: Optional[str]
+    model_ability: Optional[List[str]]
     default_model_config: Optional[Dict[str, Any]]
     default_transcription_config: Optional[Dict[str, Any]]
     engine: Optional[str]

xinference/model/audio/model_spec.json

@@ -4,7 +4,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-tiny",
     "model_revision": "167c219b21f11ef214220b8fdb7536b8a88c2475",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -12,7 +12,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-tiny.en",
     "model_revision": "87c7102498dcde7456f24cfd30239ca606ed9063",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -20,7 +20,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-base",
     "model_revision": "8c1db9b51951100007a96a525d83a8ec81b3c237",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -28,7 +28,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-base.en",
     "model_revision": "911407f4214e0e1d82085af863093ec0b66f9cd6",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -36,7 +36,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-small",
     "model_revision": "998cb1a777c20db53d6033a61b977ed4c3792cac",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -44,7 +44,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-small.en",
     "model_revision": "e8727524f962ee844a7319d92be39ac1bd25655a",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -52,7 +52,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-medium",
     "model_revision": "16688beb1294bedd0a6f5cd86fe7eec57bce41ed",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -60,7 +60,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-medium.en",
     "model_revision": "2e98eb6279edf5095af0c8dedb36bdec0acd172b",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -68,7 +68,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-large-v3",
     "model_revision": "6cdf07a7e3ec3806e5d55f787915b85d4cd020b1",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -76,7 +76,7 @@
     "model_family": "whisper",
     "model_id": "openai/whisper-large-v3-turbo",
     "model_revision": "41f01f3fe87f28c78e2fbf8b568835947dd65ed9",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -84,7 +84,7 @@
     "model_family": "whisper",
     "model_id": "BELLE-2/Belle-distilwhisper-large-v2-zh",
     "model_revision": "ed25d13498fa5bac758b2fc479435b698532dfe8",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -92,7 +92,7 @@
     "model_family": "whisper",
     "model_id": "BELLE-2/Belle-whisper-large-v2-zh",
     "model_revision": "ec5bd5d78598545b7585814edde86dac2002b5b9",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -100,14 +100,14 @@
     "model_family": "whisper",
     "model_id": "BELLE-2/Belle-whisper-large-v3-zh",
     "model_revision": "3bebc7247696b39f5ab9ed22db426943ac33f600",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
     "model_name": "whisper-tiny-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-tiny",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "engine": "mlx"
   },
@@ -115,7 +115,7 @@
     "model_name": "whisper-tiny.en-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-tiny.en-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false,
     "engine": "mlx"
   },
@@ -123,7 +123,7 @@
     "model_name": "whisper-base-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-base-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "engine": "mlx"
   },
@@ -131,7 +131,7 @@
     "model_name": "whisper-base.en-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-base.en-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false,
     "engine": "mlx"
   },
@@ -139,7 +139,7 @@
     "model_name": "whisper-small-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-small-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "engine": "mlx"
   },
@@ -147,7 +147,7 @@
     "model_name": "whisper-small.en-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-small.en-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false,
     "engine": "mlx"
   },
@@ -155,7 +155,7 @@
     "model_name": "whisper-medium-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-medium-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "engine": "mlx"
   },
@@ -163,7 +163,7 @@
     "model_name": "whisper-medium.en-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-medium.en-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false,
     "engine": "mlx"
   },
@@ -171,7 +171,7 @@
     "model_name": "whisper-large-v3-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-large-v3-mlx",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "engine": "mlx"
   },
@@ -179,7 +179,7 @@
     "model_name": "whisper-large-v3-turbo-mlx",
     "model_family": "whisper",
     "model_id": "mlx-community/whisper-large-v3-turbo",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "engine": "mlx"
   },
@@ -188,7 +188,7 @@
     "model_family": "funasr",
     "model_id": "FunAudioLLM/SenseVoiceSmall",
     "model_revision": "3eb3b4eeffc2f2dde6051b853983753db33e35c3",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "default_model_config": {
       "vad_model": "fsmn-vad",
@@ -208,7 +208,7 @@
     "model_family": "funasr",
     "model_id": "funasr/paraformer-zh",
     "model_revision": "5ed094cdfc8f6a9b6b022bd08bc904ef862bc79e",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false,
     "default_model_config": {
       "vad_model": "fsmn-vad",
@@ -223,7 +223,7 @@
     "model_family": "ChatTTS",
     "model_id": "2Noise/ChatTTS",
     "model_revision": "1a3c04a8b0651689bd9242fbb55b1f4b5a9aef84",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -231,7 +231,7 @@
     "model_family": "CosyVoice",
     "model_id": "FunAudioLLM/CosyVoice-300M",
     "model_revision": "39c4e13d46bd4dfb840d214547623e5fcd2428e2",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -239,7 +239,7 @@
     "model_family": "CosyVoice",
     "model_id": "FunAudioLLM/CosyVoice-300M-SFT",
     "model_revision": "096a5cff8d497fabb3dec2756a200f3688457a1b",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -247,7 +247,7 @@
     "model_family": "CosyVoice",
     "model_id": "FunAudioLLM/CosyVoice-300M-Instruct",
     "model_revision": "ba5265d9a3169c1fedce145122c9dd4bc24e062c",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -255,7 +255,7 @@
     "model_family": "CosyVoice",
     "model_id": "mrfakename/CosyVoice2-0.5B",
     "model_revision": "5676baabc8a76dc93ef60a88bbd2420deaa2f644",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -263,7 +263,7 @@
     "model_family": "FishAudio",
     "model_id": "fishaudio/fish-speech-1.5",
     "model_revision": "268b6ec86243dd683bc78dab7e9a6cedf9191f2a",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -271,7 +271,7 @@
     "model_family": "F5-TTS",
     "model_id": "SWivid/F5-TTS",
     "model_revision": "4dcc16f297f2ff98a17b3726b16f5de5a5e45672",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -279,7 +279,7 @@
     "model_family": "F5-TTS-MLX",
     "model_id": "lucasnewman/f5-tts-mlx",
     "model_revision": "7642bb232e3fcacf92c51c786edebb8624da6b93",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -287,7 +287,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-English",
     "model_revision": "bb4fb7346d566d277ba8c8c7dbfdf6786139b8ef",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "EN"
   },
@@ -296,7 +296,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-English-v2",
     "model_revision": "a53e3509c4ee4ff16d79272feb2474ff864e18f3",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "EN"
   },
@@ -305,7 +305,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-English-v3",
     "model_revision": "f7c4a35392c0e9be24a755f1edb4c3f63040f759",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "EN"
   },
@@ -314,7 +314,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-French",
     "model_revision": "1e9bf590262392d8bffb679b0a3b0c16b0f9fdaf",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "FR"
   },
@@ -323,7 +323,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-Japanese",
     "model_revision": "367f8795464b531b4e97c1515bddfc1243e60891",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "JP"
   },
@@ -332,7 +332,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-Spanish",
     "model_revision": "dbb5496df39d11a66c1d5f5a9ca357c3c9fb95fb",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "ES"
   },
@@ -341,7 +341,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-Chinese",
     "model_revision": "af5d207a364ea4208c6f589c89f57f88414bdd16",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "ZH"
   },
@@ -350,7 +350,7 @@
     "model_family": "MeloTTS",
     "model_id": "myshell-ai/MeloTTS-Korean",
     "model_revision": "0207e5adfc90129a51b6b03d89be6d84360ed323",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": false,
     "language": "KR"
   },
@@ -359,7 +359,7 @@
     "model_family": "Kokoro",
     "model_id": "hexgrad/Kokoro-82M",
     "model_revision": "7884269d6fd3f9beabc271b6f1308e5699281fa9",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -367,7 +367,7 @@
     "model_family": "MegaTTS",
     "model_id": "ByteDance/MegaTTS3",
     "model_revision": "409a7002b006d80f0730fca6f80441b08c10e738",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   }
 ]

xinference/model/audio/model_spec_modelscope.json

@@ -5,7 +5,7 @@
     "model_hub": "modelscope",
     "model_id": "AI-ModelScope/whisper-large-v3",
     "model_revision": "master",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -14,7 +14,7 @@
     "model_hub": "modelscope",
     "model_id": "AI-ModelScope/whisper-large-v3-turbo",
     "model_revision": "master",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true
   },
   {
@@ -23,7 +23,7 @@
     "model_hub": "modelscope",
     "model_id": "Xorbits/Belle-whisper-large-v3-zh",
     "model_revision": "master",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false
   },
   {
@@ -32,7 +32,7 @@
     "model_hub": "modelscope",
     "model_id": "iic/SenseVoiceSmall",
     "model_revision": "master",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": true,
     "default_model_config": {
       "vad_model": "fsmn-vad",
@@ -53,7 +53,7 @@
     "model_hub": "modelscope",
     "model_id": "iic/speech_paraformer-large-vad-punc-spk_asr_nat-zh-cn",
     "model_revision": "master",
-    "model_ability": "audio-to-text",
+    "model_ability": ["audio2text"],
     "multilingual": false,
     "default_model_config": {
       "vad_model": "fsmn-vad",
@@ -69,7 +69,7 @@
     "model_hub": "modelscope",
     "model_id": "AI-ModelScope/ChatTTS",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -78,7 +78,7 @@
     "model_hub": "modelscope",
     "model_id": "iic/CosyVoice-300M",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -87,7 +87,7 @@
     "model_hub": "modelscope",
     "model_id": "iic/CosyVoice-300M-SFT",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -96,7 +96,7 @@
     "model_hub": "modelscope",
     "model_id": "iic/CosyVoice-300M-Instruct",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -105,7 +105,7 @@
     "model_hub": "modelscope",
     "model_id": "iic/CosyVoice2-0.5B",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -114,7 +114,7 @@
     "model_hub": "modelscope",
     "model_id": "SWivid/F5-TTS_Emilia-ZH-EN",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -123,7 +123,7 @@
     "model_hub": "modelscope",
     "model_id": "AI-ModelScope/Kokoro-82M",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   },
   {
@@ -132,7 +132,7 @@
     "model_hub": "modelscope",
     "model_id": "ByteDance/MegaTTS3",
     "model_revision": "master",
-    "model_ability": "text-to-audio",
+    "model_ability": ["text2audio"],
     "multilingual": true
   }
 ]