Added 3 new Replicate models

.github/opencode-memory/features.md

@@ -33,3 +33,17 @@ Repository configured with OpenCode memory and automated sync workflow.
 - Model ID: `google/veo-3.1`
 - Namespace: `video.generate`
 - Description: Google's improved video model with higher-fidelity video, context-aware audio, reference image and last frame support
+
+## 2026-01-20 - New Featured Models Added
+
+- Model ID: `black-forest-labs/flux-2-klein-4b`
+- Namespace: `image.generate`
+- Description: FLUX.2 Klein - Very fast image generation model, 4 steps distilled, sub-second inference for production and near real-time applications
+
+- Model ID: `black-forest-labs/flux-2-max`
+- Namespace: `image.generate`
+- Description: FLUX.2 Max - The highest fidelity image model from Black Forest Labs
+
+- Model ID: `bytedance/seedance-1.5-pro`
+- Namespace: `video.generate`
+- Description: Seedance 1.5 Pro - Joint audio-video model that accurately follows complex instructions

src/nodetool/dsl/replicate/audio/enhance.py

@@ -1,28 +1,51 @@
+# This file is auto-generated by nodetool.dsl.codegen.
+# Please do not edit this file manually.
+
+# Instead, edit the node class in the source module and run the following commands to regenerate the DSL:
+# nodetool package scan
+# nodetool codegen
+
 from pydantic import BaseModel, Field
 import typing
 from typing import Any
 import nodetool.metadata.types
 import nodetool.metadata.types as types
-from nodetool.dsl.graph import GraphNode
+from nodetool.dsl.graph import GraphNode, SingleOutputGraphNode
+
+import typing
+from pydantic import Field
+from nodetool.dsl.handles import OutputHandle, OutputsProxy, connect_field
+import nodetool.nodes.replicate.audio.enhance
+from nodetool.workflows.base_node import BaseNode
 
 
-class AudioSuperResolution(GraphNode):
-    """AudioSR: Versatile Audio Super-resolution at Scale"""
+class AudioSuperResolution(
+    SingleOutputGraphNode[types.AudioRef], GraphNode[types.AudioRef]
+):
+    """
+    AudioSR: Versatile Audio Super-resolution at Scale
+    """
 
-    seed: int | None | GraphNode | tuple[GraphNode, str] = Field(
+    seed: int | OutputHandle[int] | None = connect_field(
         default=None, description="Random seed. Leave blank to randomize the seed"
     )
-    ddim_steps: int | GraphNode | tuple[GraphNode, str] = Field(
+    ddim_steps: int | OutputHandle[int] = connect_field(
         default=50, description="Number of inference steps"
     )
-    input_file: types.AudioRef | GraphNode | tuple[GraphNode, str] = Field(
-        default=types.AudioRef(type="audio", uri="", asset_id=None, data=None),
+    input_file: types.AudioRef | OutputHandle[types.AudioRef] = connect_field(
+        default=types.AudioRef(
+            type="audio", uri="", asset_id=None, data=None, metadata=None
+        ),
         description="Audio to upsample",
     )
-    guidance_scale: float | GraphNode | tuple[GraphNode, str] = Field(
+    guidance_scale: float | OutputHandle[float] = connect_field(
         default=3.5, description="Scale for classifier free guidance"
     )
 
+    @classmethod
+    def get_node_class(cls) -> type[BaseNode]:
+        return nodetool.nodes.replicate.audio.enhance.AudioSuperResolution
+
     @classmethod
     def get_node_type(cls):
-        return "replicate.audio.enhance.AudioSuperResolution"
+        return cls.get_node_class().get_node_type()

src/nodetool/dsl/replicate/audio/separate.py

@@ -1,20 +1,28 @@
+# This file is auto-generated by nodetool.dsl.codegen.
+# Please do not edit this file manually.
+
+# Instead, edit the node class in the source module and run the following commands to regenerate the DSL:
+# nodetool package scan
+# nodetool codegen
+
 from pydantic import BaseModel, Field
 import typing
 from typing import Any
 import nodetool.metadata.types
 import nodetool.metadata.types as types
-from nodetool.dsl.graph import GraphNode
+from nodetool.dsl.graph import GraphNode, SingleOutputGraphNode
 
+import typing
+from pydantic import Field
+from nodetool.dsl.handles import OutputHandle, OutputsProxy, connect_field
 import nodetool.nodes.replicate.audio.separate
-import nodetool.nodes.replicate.audio.separate
-import nodetool.nodes.replicate.audio.separate
-import nodetool.nodes.replicate.audio.separate
-import nodetool.nodes.replicate.audio.separate
-import nodetool.nodes.replicate.audio.separate
+from nodetool.workflows.base_node import BaseNode
 
 
-class Demucs(GraphNode):
-    """Demucs is an audio source separator created by Facebook Research."""
+class Demucs(SingleOutputGraphNode[typing.Any], GraphNode[typing.Any]):
+    """
+    Demucs is an audio source separator created by Facebook Research.
+    """
 
     Stem: typing.ClassVar[type] = nodetool.nodes.replicate.audio.separate.Demucs.Stem
     Model: typing.ClassVar[type] = nodetool.nodes.replicate.audio.separate.Demucs.Model
@@ -30,58 +38,65 @@ class Demucs(GraphNode):
     Output_format: typing.ClassVar[type] = (
         nodetool.nodes.replicate.audio.separate.Demucs.Output_format
     )
-    jobs: int | GraphNode | tuple[GraphNode, str] = Field(
+
+    jobs: int | OutputHandle[int] = connect_field(
         default=0,
         description="Choose the number of parallel jobs to use for separation.",
     )
     stem: nodetool.nodes.replicate.audio.separate.Demucs.Stem = Field(
-        default=nodetool.nodes.replicate.audio.separate.Demucs.Stem.NONE,
+        default=nodetool.nodes.replicate.audio.separate.Demucs.Stem("none"),
         description="If you just want to isolate one stem, you can choose it here.",
     )
-    audio: types.AudioRef | GraphNode | tuple[GraphNode, str] = Field(
-        default=types.AudioRef(type="audio", uri="", asset_id=None, data=None),
+    audio: types.AudioRef | OutputHandle[types.AudioRef] = connect_field(
+        default=types.AudioRef(
+            type="audio", uri="", asset_id=None, data=None, metadata=None
+        ),
         description="Upload the file to be processed here.",
     )
     model: nodetool.nodes.replicate.audio.separate.Demucs.Model = Field(
-        default=nodetool.nodes.replicate.audio.separate.Demucs.Model.HTDEMUCS,
+        default=nodetool.nodes.replicate.audio.separate.Demucs.Model("htdemucs"),
         description="Choose the demucs audio that proccesses your audio. The readme has more information on what to choose.",
     )
-    split: bool | GraphNode | tuple[GraphNode, str] = Field(
+    split: bool | OutputHandle[bool] = connect_field(
         default=True,
         description="Choose whether or not the audio should be split into chunks.",
     )
-    shifts: int | GraphNode | tuple[GraphNode, str] = Field(
+    shifts: int | OutputHandle[int] = connect_field(
         default=1,
         description="Choose the amount random shifts for equivariant stabilization. This performs multiple predictions with random shifts of the input and averages them, which makes it x times slower.",
     )
-    overlap: float | GraphNode | tuple[GraphNode, str] = Field(
+    overlap: float | OutputHandle[float] = connect_field(
         default=0.25,
         description="Choose the amount of overlap between prediction windows.",
     )
-    segment: int | None | GraphNode | tuple[GraphNode, str] = Field(
+    segment: int | OutputHandle[int] | None = connect_field(
         default=None, description="Choose the segment length to use for separation."
     )
     clip_mode: nodetool.nodes.replicate.audio.separate.Demucs.Clip_mode = Field(
-        default=nodetool.nodes.replicate.audio.separate.Demucs.Clip_mode.RESCALE,
+        default=nodetool.nodes.replicate.audio.separate.Demucs.Clip_mode("rescale"),
         description="Choose the strategy for avoiding clipping. Rescale will rescale entire signal if necessary or clamp will allow hard clipping.",
     )
     mp3_preset: nodetool.nodes.replicate.audio.separate.Demucs.Mp3_preset = Field(
-        default=nodetool.nodes.replicate.audio.separate.Demucs.Mp3_preset._2,
+        default=nodetool.nodes.replicate.audio.separate.Demucs.Mp3_preset(2),
         description="Choose the preset for the MP3 output. Higher is faster but worse quality. If MP3 is not selected as the output type, this has no effect.",
     )
     wav_format: nodetool.nodes.replicate.audio.separate.Demucs.Wav_format = Field(
-        default=nodetool.nodes.replicate.audio.separate.Demucs.Wav_format.INT24,
+        default=nodetool.nodes.replicate.audio.separate.Demucs.Wav_format("int24"),
         description="Choose format for the WAV output. If WAV is not selected as the output type, this has no effect.",
     )
-    mp3_bitrate: int | GraphNode | tuple[GraphNode, str] = Field(
+    mp3_bitrate: int | OutputHandle[int] = connect_field(
         default=320,
         description="Choose the bitrate for the MP3 output. Higher is better quality but larger file size. If MP3 is not selected as the output type, this has no effect.",
     )
     output_format: nodetool.nodes.replicate.audio.separate.Demucs.Output_format = Field(
-        default=nodetool.nodes.replicate.audio.separate.Demucs.Output_format.MP3,
+        default=nodetool.nodes.replicate.audio.separate.Demucs.Output_format("mp3"),
         description="Choose the audio format you would like the result to be returned in.",
     )
 
+    @classmethod
+    def get_node_class(cls) -> type[BaseNode]:
+        return nodetool.nodes.replicate.audio.separate.Demucs
+
     @classmethod
     def get_node_type(cls):
-        return "replicate.audio.separate.Demucs"
+        return cls.get_node_class().get_node_type()

src/nodetool/dsl/replicate/audio/transcribe.py

@@ -1,17 +1,65 @@
+# This file is auto-generated by nodetool.dsl.codegen.
+# Please do not edit this file manually.
+
+# Instead, edit the node class in the source module and run the following commands to regenerate the DSL:
+# nodetool package scan
+# nodetool codegen
+
 from pydantic import BaseModel, Field
 import typing
 from typing import Any
 import nodetool.metadata.types
 import nodetool.metadata.types as types
-from nodetool.dsl.graph import GraphNode
+from nodetool.dsl.graph import GraphNode, SingleOutputGraphNode
 
+import typing
+from pydantic import Field
+from nodetool.dsl.handles import OutputHandle, OutputsProxy, connect_field
 import nodetool.nodes.replicate.audio.transcribe
+from nodetool.workflows.base_node import BaseNode
+
+
+class GPT4o_Transcribe(SingleOutputGraphNode[str], GraphNode[str]):
+    """
+    A speech-to-text model that uses GPT-4o to transcribe audio
+    """
+
+    prompt: str | OutputHandle[str] | None = connect_field(
+        default=None,
+        description="An optional text to guide the model's style or continue a previous audio segment. The prompt should match the audio language.",
+    )
+    language: str | OutputHandle[str] | None = connect_field(
+        default=None,
+        description="The language of the input audio. Supplying the input language in ISO-639-1 (e.g. en) format will improve accuracy and latency.",
+    )
+    audio_file: str | OutputHandle[str] | None = connect_field(
+        default=None,
+        description="The audio file to transcribe. Supported formats: mp3, mp4, mpeg, mpga, m4a, ogg, wav, or webm",
+    )
+    temperature: float | OutputHandle[float] = connect_field(
+        default=0, description="Sampling temperature between 0 and 1"
+    )
+
+    @classmethod
+    def get_node_class(cls) -> type[BaseNode]:
+        return nodetool.nodes.replicate.audio.transcribe.GPT4o_Transcribe
+
+    @classmethod
+    def get_node_type(cls):
+        return cls.get_node_class().get_node_type()
+
+
+import typing
+from pydantic import Field
+from nodetool.dsl.handles import OutputHandle, OutputsProxy, connect_field
 import nodetool.nodes.replicate.audio.transcribe
-import nodetool.nodes.replicate.audio.transcribe
+from nodetool.workflows.base_node import BaseNode
 
 
-class IncrediblyFastWhisper(GraphNode):
-    """whisper-large-v3, incredibly fast, powered by Hugging Face Transformers! 🤗"""
+class IncrediblyFastWhisper(SingleOutputGraphNode[str], GraphNode[str]):
+    """
+    whisper-large-v3, incredibly fast, powered by Hugging Face Transformers! 🤗
+    """
 
     Task: typing.ClassVar[type] = (
         nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Task
@@ -22,39 +70,52 @@ class IncrediblyFastWhisper(GraphNode):
     Timestamp: typing.ClassVar[type] = (
         nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Timestamp
     )
+
     task: nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Task = Field(
-        default=nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Task.TRANSCRIBE,
+        default=nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Task(
+            "transcribe"
+        ),
         description="Task to perform: transcribe or translate to another language.",
     )
-    audio: types.AudioRef | GraphNode | tuple[GraphNode, str] = Field(
-        default=types.AudioRef(type="audio", uri="", asset_id=None, data=None),
+    audio: types.AudioRef | OutputHandle[types.AudioRef] = connect_field(
+        default=types.AudioRef(
+            type="audio", uri="", asset_id=None, data=None, metadata=None
+        ),
         description="Audio file",
     )
-    hf_token: str | None | GraphNode | tuple[GraphNode, str] = Field(
+    hf_token: str | OutputHandle[str] | None = connect_field(
         default=None,
         description="Provide a hf.co/settings/token for Pyannote.audio to diarise the audio clips. You need to agree to the terms in 'https://huggingface.co/pyannote/speaker-diarization-3.1' and 'https://huggingface.co/pyannote/segmentation-3.0' first.",
     )
     language: (
         nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Language
     ) = Field(
-        default=nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Language.NONE,
+        default=nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Language(
+            "None"
+        ),
         description="Language spoken in the audio, specify 'None' to perform language detection.",
     )
     timestamp: (
         nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Timestamp
     ) = Field(
-        default=nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Timestamp.CHUNK,
+        default=nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper.Timestamp(
+            "chunk"
+        ),
         description="Whisper supports both chunked as well as word level timestamps.",
     )
-    batch_size: int | GraphNode | tuple[GraphNode, str] = Field(
+    batch_size: int | OutputHandle[int] = connect_field(
         default=24,
         description="Number of parallel batches you want to compute. Reduce if you face OOMs.",
     )
-    diarise_audio: bool | GraphNode | tuple[GraphNode, str] = Field(
+    diarise_audio: bool | OutputHandle[bool] = connect_field(
         default=False,
         description="Use Pyannote.audio to diarise the audio clips. You will need to provide hf_token below too.",
     )
 
+    @classmethod
+    def get_node_class(cls) -> type[BaseNode]:
+        return nodetool.nodes.replicate.audio.transcribe.IncrediblyFastWhisper
+
     @classmethod
     def get_node_type(cls):
-        return "replicate.audio.transcribe.IncrediblyFastWhisper"
+        return cls.get_node_class().get_node_type()