release: 2.16.0

.release-please-manifest.json

@@ -1,3 +1,3 @@
 {
-  ".": "2.15.0"
+  ".": "2.16.0"
 }

.stats.yml

@@ -1,4 +1,4 @@
 configured_endpoints: 81
-openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/togetherai/togetherai-ce108a2095d36552bb556506de04475674f512a13bc5aa099e9750993405be14.yml
-openapi_spec_hash: 4763dd426dd805306bbb38a314158cd3
+openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/togetherai/togetherai-891d6a19c1efb39d010d53a8965fd314acf279fcb09f7a6622df5fc6ce779bec.yml
+openapi_spec_hash: 4554c2095bf588ac4a195d038f3893c1
 config_hash: b35d5968fb07cce1c1be735f874898b1

CHANGELOG.md

@@ -1,5 +1,18 @@
 # Changelog
 
+## 2.16.0 (2026-05-22)
+
+Full Changelog: [v2.15.0...v2.16.0](https://github.com/togethercomputer/together-py/compare/v2.15.0...v2.16.0)
+
+### Features
+
+* **cli:** expose new cluster SDK parameters ([#378](https://github.com/togethercomputer/together-py/issues/378)) ([b694d8f](https://github.com/togethercomputer/together-py/commit/b694d8f261da15835257bb0352e1ba312b4127d3))
+
+
+### Documentation
+
+* **api:** add size and duration limits to file parameter in audio transcriptions/translations ([cce54f2](https://github.com/togethercomputer/together-py/commit/cce54f2362fd8dba1205eccf89f6909be74e2c37))
+
 ## 2.15.0 (2026-05-20)
 
 Full Changelog: [v2.14.0...v2.15.0](https://github.com/togethercomputer/together-py/compare/v2.14.0...v2.15.0)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "together"
-version = "2.15.0"
+version = "2.16.0"
 description = "The official Python library for the together API"
 dynamic = ["readme"]
 license = "Apache-2.0"

src/together/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "together"
-__version__ = "2.15.0"  # x-release-please-version
+__version__ = "2.16.0"  # x-release-please-version

src/together/lib/cli/api/beta/clusters/create.py

@@ -14,7 +14,8 @@
 NumGpusParameter = Annotated[Optional[int], Parameter(help="Number of GPUs to allocate in the cluster")]
 RegionParameter = Annotated[Optional[str], Parameter(help="Region to create the cluster in")]
 BillingTypeParameter = Annotated[
-    Optional[Literal["RESERVED", "ON_DEMAND"]], Parameter(help="Billing type to use for the cluster")
+    Optional[Literal["RESERVED", "ON_DEMAND", "SCHEDULED_CAPACITY"]],
+    Parameter(help="Billing type to use for the cluster"),
 ]
 NvidiaDriverVersionParameter = Annotated[Optional[str], Parameter(help="Nvidia driver version to use for the cluster")]
 CudaVersionParameter = Annotated[Optional[str], Parameter(help="CUDA version to use for the cluster")]
@@ -27,6 +28,25 @@
 ]
 ClusterTypeParameter = Annotated[Optional[Literal["KUBERNETES", "SLURM"]], Parameter(help="Cluster type")]
 VolumeParameter = Annotated[Optional[str], Parameter(help="Storage volume ID to use for the cluster")]
+AutoScaleParameter = Annotated[Optional[bool], Parameter(help="Enable cluster auto-scaling")]
+AutoScaleMaxGpusParameter = Annotated[Optional[int], Parameter(help="Maximum GPUs for auto-scaling")]
+CapacityPoolIDParameter = Annotated[Optional[str], Parameter(help="Capacity pool ID to use for the cluster")]
+GpuNodeFailoverEnabledParameter = Annotated[
+    Optional[bool], Parameter(help="Enable automated GPU node failover for the cluster")
+]
+InstallTraefikParameter = Annotated[Optional[bool], Parameter(help="Install Traefik ingress controller")]
+NumCapacityPoolGpusParameter = Annotated[
+    Optional[int], Parameter(help="Number of GPUs to allocate from a capacity pool")
+]
+NumPreemptibleGpusParameter = Annotated[Optional[int], Parameter(help="Number of preemptible GPUs to request")]
+NumReservedGpusParameter = Annotated[Optional[int], Parameter(help="Number of prepaid reserved GPUs to request")]
+ProjectIDParameter = Annotated[Optional[str], Parameter(help="Project ID for the cluster")]
+ReservationEndTimeParameter = Annotated[Optional[str], Parameter(help="Reservation end time for scheduled capacity")]
+ReservationStartTimeParameter = Annotated[
+    Optional[str], Parameter(help="Reservation start time for scheduled capacity")
+]
+SlurmImageParameter = Annotated[Optional[str], Parameter(help="Custom Slurm image for Slurm clusters")]
+SlurmShmSizeGibParameter = Annotated[Optional[int], Parameter(help="Shared memory size in GiB for Slurm clusters")]
 
 
 async def create(
@@ -40,6 +60,19 @@ async def create(
     gpu_type: GpuTypeParameter = None,
     cluster_type: ClusterTypeParameter = None,
     volume: VolumeParameter = None,
+    auto_scale: AutoScaleParameter = None,
+    auto_scale_max_gpus: AutoScaleMaxGpusParameter = None,
+    capacity_pool_id: CapacityPoolIDParameter = None,
+    gpu_node_failover_enabled: GpuNodeFailoverEnabledParameter = None,
+    install_traefik: InstallTraefikParameter = None,
+    num_capacity_pool_gpus: NumCapacityPoolGpusParameter = None,
+    num_preemptible_gpus: NumPreemptibleGpusParameter = None,
+    num_reserved_gpus: NumReservedGpusParameter = None,
+    project_id: ProjectIDParameter = None,
+    reservation_end_time: ReservationEndTimeParameter = None,
+    reservation_start_time: ReservationStartTimeParameter = None,
+    slurm_image: SlurmImageParameter = None,
+    slurm_shm_size_gib: SlurmShmSizeGibParameter = None,
     *,
     config: CLIConfigParameter,
 ) -> None:
@@ -57,6 +90,32 @@ async def create(
     )
     if volume:
         params["volume_id"] = volume
+    if auto_scale is not None:
+        params["auto_scale"] = auto_scale
+    if auto_scale_max_gpus is not None:
+        params["auto_scale_max_gpus"] = auto_scale_max_gpus
+    if capacity_pool_id:
+        params["capacity_pool_id"] = capacity_pool_id
+    if gpu_node_failover_enabled is not None:
+        params["gpu_node_failover_enabled"] = gpu_node_failover_enabled
+    if install_traefik is not None:
+        params["install_traefik"] = install_traefik
+    if num_capacity_pool_gpus is not None:
+        params["num_capacity_pool_gpus"] = num_capacity_pool_gpus
+    if num_preemptible_gpus is not None:
+        params["num_preemptible_gpus"] = num_preemptible_gpus
+    if num_reserved_gpus is not None:
+        params["num_reserved_gpus"] = num_reserved_gpus
+    if project_id:
+        params["project_id"] = project_id
+    if reservation_end_time:
+        params["reservation_end_time"] = reservation_end_time
+    if reservation_start_time:
+        params["reservation_start_time"] = reservation_start_time
+    if slurm_image:
+        params["slurm_image"] = slurm_image
+    if slurm_shm_size_gib is not None:
+        params["slurm_shm_size_gib"] = slurm_shm_size_gib
 
     # JSON Mode skips hand holding through the argument setup
     if not config.json and not config.non_interactive:

src/together/lib/cli/api/beta/clusters/storage/create.py

@@ -1,9 +1,10 @@
 from __future__ import annotations
 
-from typing import Annotated
+from typing import Optional, Annotated
 
 from cyclopts import Parameter
 
+from together import omit
 from together._utils._json import openapi_dumps
 from together.lib.cli.utils.config import CLIConfigParameter
 from together.lib.cli.utils._console import console
@@ -13,6 +14,10 @@ async def create(
     region: Annotated[str, Parameter(help="Region to create the storage volume in")],
     size_tib: Annotated[int, Parameter(help="Size of the storage volume in TiB")],
     volume_name: Annotated[str, Parameter(help="Name of the storage volume")],
+    is_lifecycle_independent: Annotated[
+        Optional[bool],
+        Parameter(help="Keep the storage volume after cluster decommissioning"),
+    ] = None,
     *,
     config: CLIConfigParameter,
 ) -> None:
@@ -21,6 +26,7 @@ async def create(
         region=region,
         size_tib=size_tib,
         volume_name=volume_name,
+        is_lifecycle_independent=is_lifecycle_independent if is_lifecycle_independent is not None else omit,
     )
 
     if config.json:

src/together/lib/cli/api/beta/clusters/storage/update.py

@@ -1,24 +1,25 @@
 from __future__ import annotations
 
-from typing import Annotated
+from typing import Optional, Annotated
 
 from cyclopts import Parameter
 
+from together import omit
 from together._utils._json import openapi_dumps
 from together.lib.cli.utils.config import CLIConfigParameter
 from together.lib.cli.utils._console import console
 
 
 async def update(
     volume_id: str,
-    size_tib: Annotated[int, Parameter(help="New size of the storage volume in TiB")],
+    size_tib: Annotated[Optional[int], Parameter(help="New size of the storage volume in TiB")] = None,
     *,
     config: CLIConfigParameter,
 ) -> None:
     """Update a storage volume (resize)."""
     response = await config.client.beta.clusters.storage.update(
         volume_id=volume_id,
-        size_tib=size_tib,
+        size_tib=size_tib if size_tib is not None else omit,
     )
 
     if config.json:

src/together/lib/cli/api/beta/clusters/update.py

@@ -18,6 +18,18 @@ async def update(
     cluster_type: Annotated[
         Optional[Literal["KUBERNETES", "SLURM"]], Parameter(help="Type of cluster to update")
     ] = None,
+    num_preemptible_gpus: Annotated[
+        Optional[int],
+        Parameter(help="Desired number of preemptible GPUs for the cluster"),
+    ] = None,
+    num_reserved_gpus: Annotated[
+        Optional[int],
+        Parameter(help="Desired number of reserved GPUs for the cluster"),
+    ] = None,
+    reservation_end_time: Annotated[
+        Optional[str],
+        Parameter(help="Timestamp at which the cluster should be decommissioned"),
+    ] = None,
     *,
     config: CLIConfigParameter,
 ) -> None:
@@ -29,6 +41,9 @@ async def update(
             cluster_id,
             num_gpus=num_gpus if num_gpus is not None else omit,
             cluster_type=cluster_type if cluster_type is not None else omit,
+            num_preemptible_gpus=num_preemptible_gpus if num_preemptible_gpus is not None else omit,
+            num_reserved_gpus=num_reserved_gpus if num_reserved_gpus is not None else omit,
+            reservation_end_time=reservation_end_time or omit,
         ),
     )
 

src/together/resources/audio/transcriptions.py

@@ -69,8 +69,10 @@ def create(
         Transcribes audio into text
 
         Args:
-          file: Audio file upload or public HTTP/HTTPS URL. Supported formats .wav, .mp3, .m4a,
-              .webm, .flac, .ogg, .opus, .aac.
+          file: Audio file upload or public HTTP/HTTPS URL. Supported formats: .wav, .mp3, .m4a,
+              .webm, .flac, .ogg, .opus, .aac. Maximum duration 4 hours; longer audio is
+              rejected with `audio_too_long`. Binary uploads are additionally capped at 500 MB
+              (HTTP 413); URL-fetched audio is capped at 1 GB.
 
           diarize: Whether to enable speaker diarization. When enabled, you will get the speaker id
               for each word in the transcription. In the response, in the words array, you
@@ -195,8 +197,10 @@ async def create(
         Transcribes audio into text
 
         Args:
-          file: Audio file upload or public HTTP/HTTPS URL. Supported formats .wav, .mp3, .m4a,
-              .webm, .flac, .ogg, .opus, .aac.
+          file: Audio file upload or public HTTP/HTTPS URL. Supported formats: .wav, .mp3, .m4a,
+              .webm, .flac, .ogg, .opus, .aac. Maximum duration 4 hours; longer audio is
+              rejected with `audio_too_long`. Binary uploads are additionally capped at 500 MB
+              (HTTP 413); URL-fetched audio is capped at 1 GB.
 
           diarize: Whether to enable speaker diarization. When enabled, you will get the speaker id
               for each word in the transcription. In the response, in the words array, you

src/together/resources/audio/translations.py

@@ -66,8 +66,10 @@ def create(
         Translates audio into English
 
         Args:
-          file: Audio file upload or public HTTP/HTTPS URL. Supported formats .wav, .mp3, .m4a,
-              .webm, .flac, .ogg, .opus, .aac.
+          file: Audio file upload or public HTTP/HTTPS URL. Supported formats: .wav, .mp3, .m4a,
+              .webm, .flac, .ogg, .opus, .aac. Maximum duration 4 hours; longer audio is
+              rejected with `audio_too_long`. Binary uploads are additionally capped at 500 MB
+              (HTTP 413); URL-fetched audio is capped at 1 GB.
 
           language: Target output language. Optional ISO 639-1 language code. If omitted, language
               is set to English.
@@ -169,8 +171,10 @@ async def create(
         Translates audio into English
 
         Args:
-          file: Audio file upload or public HTTP/HTTPS URL. Supported formats .wav, .mp3, .m4a,
-              .webm, .flac, .ogg, .opus, .aac.
+          file: Audio file upload or public HTTP/HTTPS URL. Supported formats: .wav, .mp3, .m4a,
+              .webm, .flac, .ogg, .opus, .aac. Maximum duration 4 hours; longer audio is
+              rejected with `audio_too_long`. Binary uploads are additionally capped at 500 MB
+              (HTTP 413); URL-fetched audio is capped at 1 GB.
 
           language: Target output language. Optional ISO 639-1 language code. If omitted, language
               is set to English.

src/together/types/audio/transcription_create_params.py

@@ -14,7 +14,10 @@ class TranscriptionCreateParams(TypedDict, total=False):
     file: Required[Union[FileTypes, str]]
     """Audio file upload or public HTTP/HTTPS URL.
 
-    Supported formats .wav, .mp3, .m4a, .webm, .flac, .ogg, .opus, .aac.
+    Supported formats: .wav, .mp3, .m4a, .webm, .flac, .ogg, .opus, .aac. Maximum
+    duration 4 hours; longer audio is rejected with `audio_too_long`. Binary uploads
+    are additionally capped at 500 MB (HTTP 413); URL-fetched audio is capped at 1
+    GB.
     """
 
     diarize: bool

src/together/types/audio/translation_create_params.py

@@ -14,7 +14,10 @@ class TranslationCreateParams(TypedDict, total=False):
     file: Required[Union[FileTypes, str]]
     """Audio file upload or public HTTP/HTTPS URL.
 
-    Supported formats .wav, .mp3, .m4a, .webm, .flac, .ogg, .opus, .aac.
+    Supported formats: .wav, .mp3, .m4a, .webm, .flac, .ogg, .opus, .aac. Maximum
+    duration 4 hours; longer audio is rejected with `audio_too_long`. Binary uploads
+    are additionally capped at 500 MB (HTTP 413); URL-fetched audio is capped at 1
+    GB.
     """
 
     language: str