diff --git a/.release-please-manifest.json b/.release-please-manifest.json
index ec9417a..68f4960 100644
--- a/.release-please-manifest.json
+++ b/.release-please-manifest.json
@@ -1,3 +1,3 @@
 {
-  ".": "2.0.0-alpha.27"
+  ".": "2.0.0-alpha.28"
 }
\ No newline at end of file
diff --git a/.stats.yml b/.stats.yml
index 0277638..627f641 100644
--- a/.stats.yml
+++ b/.stats.yml
@@ -1,4 +1,4 @@
 configured_endpoints: 36
-openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/replicate%2Freplicate-client-87c7c57bd75c54990c679c9e87d009851cdff572815a55d1b6ee4d4ee20adaa1.yml
-openapi_spec_hash: d987f14befa536004eece7b49caad993
-config_hash: a916e7f3559ab312c7b6696cd6b35fb5
+openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/replicate%2Freplicate-client-8b68f03c8602e25ede74e66221e04bb17ac1f3170b8de49b8311fbe492636245.yml
+openapi_spec_hash: 81a7c6f32c6d77c1f329ca5a3c700a7a
+config_hash: 407acf62c906ee301314f2d23cdb58b1
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 1dae4bb..1b486c4 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,24 @@
 # Changelog
 
+## 2.0.0-alpha.28 (2025-10-07)
+
+Full Changelog: [v2.0.0-alpha.27...v2.0.0-alpha.28](https://github.com/replicate/replicate-python-stainless/compare/v2.0.0-alpha.27...v2.0.0-alpha.28)
+
+### Chores
+
+* do not install brew dependencies in ./scripts/bootstrap by default ([40f38a7](https://github.com/replicate/replicate-python-stainless/commit/40f38a774220bb1364e12de4d7aef9448fb60c58))
+* **types:** change optional parameter type from NotGiven to Omit ([f331b97](https://github.com/replicate/replicate-python-stainless/commit/f331b97a15226c9ae17112567a3be01010439f21))
+* update OpenAPI spec and rebuild SDKs ([8b43277](https://github.com/replicate/replicate-python-stainless/commit/8b432772af94ac3f471bc642671322c11c7d4ff0))
+* update OpenAPI spec and rebuild SDKs ([77c9c21](https://github.com/replicate/replicate-python-stainless/commit/77c9c216dea9886a412fd5a1b8f0dfec2e0de155))
+* update OpenAPI spec and rebuild SDKs ([d5ed889](https://github.com/replicate/replicate-python-stainless/commit/d5ed889c6f84e0ea48129f12db494d1d08e3ffbf))
+* update OpenAPI spec and rebuild SDKs ([0a821f8](https://github.com/replicate/replicate-python-stainless/commit/0a821f8f63208585cd4730a50842907ce26a3139))
+* update OpenAPI spec and rebuild SDKs ([a4bdae2](https://github.com/replicate/replicate-python-stainless/commit/a4bdae264cff81657e2e7dcd43eb41bbd1f15237))
+
+
+### Documentation
+
+* remove replicate.stream from README ([#78](https://github.com/replicate/replicate-python-stainless/issues/78)) ([9efac9a](https://github.com/replicate/replicate-python-stainless/commit/9efac9a97e7db41d5829c055959b29f154f4c175))
+
 ## 2.0.0-alpha.27 (2025-09-29)
 
 Full Changelog: [v2.0.0-alpha.26...v2.0.0-alpha.27](https://github.com/replicate/replicate-python-stainless/compare/v2.0.0-alpha.26...v2.0.0-alpha.27)
diff --git a/pyproject.toml b/pyproject.toml
index 9ba41af..5137c3d 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "replicate"
-version = "2.0.0-alpha.27"
+version = "2.0.0-alpha.28"
 description = "The official Python library for the replicate API"
 dynamic = ["readme"]
 license = "Apache-2.0"
diff --git a/scripts/bootstrap b/scripts/bootstrap
index e84fe62..b430fee 100755
--- a/scripts/bootstrap
+++ b/scripts/bootstrap
@@ -4,10 +4,18 @@ set -e
 
 cd "$(dirname "$0")/.."
 
-if ! command -v rye >/dev/null 2>&1 && [ -f "Brewfile" ] && [ "$(uname -s)" = "Darwin" ]; then
+if [ -f "Brewfile" ] && [ "$(uname -s)" = "Darwin" ] && [ "$SKIP_BREW" != "1" ] && [ -t 0 ]; then
   brew bundle check >/dev/null 2>&1 || {
-    echo "==> Installing Homebrew dependencies…"
-    brew bundle
+    echo -n "==> Install Homebrew dependencies? (y/N): "
+    read -r response
+    case "$response" in
+      [yY][eE][sS]|[yY])
+        brew bundle
+        ;;
+      *)
+        ;;
+    esac
+    echo
   }
 fi
 
diff --git a/src/replicate/_client.py b/src/replicate/_client.py
index 390a552..5fb9334 100644
--- a/src/replicate/_client.py
+++ b/src/replicate/_client.py
@@ -29,7 +29,6 @@
 from ._qs import Querystring
 from .types import client_search_params
 from ._types import (
-    NOT_GIVEN,
     Body,
     Omit,
     Query,
@@ -39,6 +38,8 @@
     Transport,
     ProxiesTypes,
     RequestOptions,
+    omit,
+    not_given,
 )
 from ._utils import (
     is_given,
@@ -103,7 +104,7 @@ def __init__(
         *,
         bearer_token: str | None = None,
         base_url: str | httpx.URL | None = None,
-        timeout: Union[float, Timeout, None, NotGiven] = NOT_GIVEN,
+        timeout: float | Timeout | None | NotGiven = not_given,
         max_retries: int = DEFAULT_MAX_RETRIES,
         default_headers: Mapping[str, str] | None = None,
         default_query: Mapping[str, object] | None = None,
@@ -241,7 +242,7 @@ def run(
         *,
         file_encoding_strategy: Optional["FileEncodingStrategy"] = None,
         use_file_output: bool = True,
-        wait: Union[int, bool, NotGiven] = NOT_GIVEN,
+        wait: Union[int, bool, NotGiven] = not_given,
         **params: Unpack[PredictionCreateParamsWithoutVersion],
     ) -> Any:
         """
@@ -325,9 +326,9 @@ def copy(
         *,
         bearer_token: str | None = None,
         base_url: str | httpx.URL | None = None,
-        timeout: float | Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | Timeout | None | NotGiven = not_given,
         http_client: httpx.Client | None = None,
-        max_retries: int | NotGiven = NOT_GIVEN,
+        max_retries: int | NotGiven = not_given,
         default_headers: Mapping[str, str] | None = None,
         set_default_headers: Mapping[str, str] | None = None,
         default_query: Mapping[str, object] | None = None,
@@ -375,13 +376,13 @@ def search(
         self,
         *,
         query: str,
-        limit: int | NotGiven = NOT_GIVEN,
+        limit: int | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> SearchResponse:
         """
         Search for public models, collections, and docs using a text query.
@@ -478,7 +479,7 @@ def __init__(
         *,
         bearer_token: str | None = None,
         base_url: str | httpx.URL | None = None,
-        timeout: Union[float, Timeout, None, NotGiven] = NOT_GIVEN,
+        timeout: float | Timeout | None | NotGiven = not_given,
         max_retries: int = DEFAULT_MAX_RETRIES,
         default_headers: Mapping[str, str] | None = None,
         default_query: Mapping[str, object] | None = None,
@@ -616,7 +617,7 @@ async def run(
         *,
         use_file_output: bool = True,
         file_encoding_strategy: Optional["FileEncodingStrategy"] = None,
-        wait: Union[int, bool, NotGiven] = NOT_GIVEN,
+        wait: Union[int, bool, NotGiven] = not_given,
         **params: Unpack[PredictionCreateParamsWithoutVersion],
     ) -> Any:
         """
@@ -700,9 +701,9 @@ def copy(
         *,
         bearer_token: str | None = None,
         base_url: str | httpx.URL | None = None,
-        timeout: float | Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | Timeout | None | NotGiven = not_given,
         http_client: httpx.AsyncClient | None = None,
-        max_retries: int | NotGiven = NOT_GIVEN,
+        max_retries: int | NotGiven = not_given,
         default_headers: Mapping[str, str] | None = None,
         set_default_headers: Mapping[str, str] | None = None,
         default_query: Mapping[str, object] | None = None,
@@ -750,13 +751,13 @@ async def search(
         self,
         *,
         query: str,
-        limit: int | NotGiven = NOT_GIVEN,
+        limit: int | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> SearchResponse:
         """
         Search for public models, collections, and docs using a text query.
diff --git a/src/replicate/_version.py b/src/replicate/_version.py
index da0284e..80a4c91 100644
--- a/src/replicate/_version.py
+++ b/src/replicate/_version.py
@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "replicate"
-__version__ = "2.0.0-alpha.27"  # x-release-please-version
+__version__ = "2.0.0-alpha.28"  # x-release-please-version
diff --git a/src/replicate/resources/collections.py b/src/replicate/resources/collections.py
index 1a89c01..823257e 100644
--- a/src/replicate/resources/collections.py
+++ b/src/replicate/resources/collections.py
@@ -113,6 +113,7 @@ def get(
           "name": "Super resolution",
           "slug": "super-resolution",
           "description": "Upscaling models that create high-quality images from low-quality images.",
+          "full_description": "## Overview\n\nThese models generate high-quality images from low-quality images. Many of these models are based on **advanced upscaling techniques**.\n\n### Key Features\n\n- Enhance image resolution\n- Restore fine details\n- Improve overall image quality",
           "models": [...]
         }
         ```
@@ -229,6 +230,7 @@ async def get(
           "name": "Super resolution",
           "slug": "super-resolution",
           "description": "Upscaling models that create high-quality images from low-quality images.",
+          "full_description": "## Overview\n\nThese models generate high-quality images from low-quality images. Many of these models are based on **advanced upscaling techniques**.\n\n### Key Features\n\n- Enhance image resolution\n- Restore fine details\n- Improve overall image quality",
           "models": [...]
         }
         ```
diff --git a/src/replicate/resources/deployments/predictions.py b/src/replicate/resources/deployments/predictions.py
index 572ba33..23a487c 100644
--- a/src/replicate/resources/deployments/predictions.py
+++ b/src/replicate/resources/deployments/predictions.py
@@ -54,6 +54,7 @@ def create(
         webhook: str | Omit = omit,
         webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | Omit = omit,
         prefer: str | Omit = omit,
+        replicate_max_lifetime: str | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
@@ -111,7 +112,7 @@ def create(
               [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
               This field is no longer needed as the returned prediction will always have a
-              `stream` entry in its `url` property if the model supports streaming.
+              `stream` entry in its `urls` property if the model supports streaming.
 
           webhook: An HTTPS URL for receiving a webhook when the prediction has new output. The
               webhook will be a POST request where the request body is the same as the
@@ -162,7 +163,15 @@ def create(
             raise ValueError(f"Expected a non-empty value for `deployment_owner` but received {deployment_owner!r}")
         if not deployment_name:
             raise ValueError(f"Expected a non-empty value for `deployment_name` but received {deployment_name!r}")
-        extra_headers = {**strip_not_given({"Prefer": prefer}), **(extra_headers or {})}
+        extra_headers = {
+            **strip_not_given(
+                {
+                    "Prefer": prefer,
+                    "Replicate-Max-Lifetime": replicate_max_lifetime,
+                }
+            ),
+            **(extra_headers or {}),
+        }
         return self._post(
             f"/deployments/{deployment_owner}/{deployment_name}/predictions",
             body=maybe_transform(
@@ -211,6 +220,7 @@ async def create(
         webhook: str | Omit = omit,
         webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | Omit = omit,
         prefer: str | Omit = omit,
+        replicate_max_lifetime: str | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
@@ -268,7 +278,7 @@ async def create(
               [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
               This field is no longer needed as the returned prediction will always have a
-              `stream` entry in its `url` property if the model supports streaming.
+              `stream` entry in its `urls` property if the model supports streaming.
 
           webhook: An HTTPS URL for receiving a webhook when the prediction has new output. The
               webhook will be a POST request where the request body is the same as the
@@ -319,7 +329,15 @@ async def create(
             raise ValueError(f"Expected a non-empty value for `deployment_owner` but received {deployment_owner!r}")
         if not deployment_name:
             raise ValueError(f"Expected a non-empty value for `deployment_name` but received {deployment_name!r}")
-        extra_headers = {**strip_not_given({"Prefer": prefer}), **(extra_headers or {})}
+        extra_headers = {
+            **strip_not_given(
+                {
+                    "Prefer": prefer,
+                    "Replicate-Max-Lifetime": replicate_max_lifetime,
+                }
+            ),
+            **(extra_headers or {}),
+        }
         return await self._post(
             f"/deployments/{deployment_owner}/{deployment_name}/predictions",
             body=await async_maybe_transform(
diff --git a/src/replicate/resources/models/predictions.py b/src/replicate/resources/models/predictions.py
index f88e551..eebedfd 100644
--- a/src/replicate/resources/models/predictions.py
+++ b/src/replicate/resources/models/predictions.py
@@ -56,6 +56,7 @@ def create(
         webhook: str | Omit = omit,
         webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | Omit = omit,
         prefer: str | Omit = omit,
+        replicate_max_lifetime: str | Omit = omit,
         file_encoding_strategy: Optional["FileEncodingStrategy"] = None,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
@@ -118,7 +119,7 @@ def create(
               [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
               This field is no longer needed as the returned prediction will always have a
-              `stream` entry in its `url` property if the model supports streaming.
+              `stream` entry in its `urls` property if the model supports streaming.
 
           webhook: An HTTPS URL for receiving a webhook when the prediction has new output. The
               webhook will be a POST request where the request body is the same as the
@@ -169,7 +170,15 @@ def create(
             raise ValueError(f"Expected a non-empty value for `model_owner` but received {model_owner!r}")
         if not model_name:
             raise ValueError(f"Expected a non-empty value for `model_name` but received {model_name!r}")
-        extra_headers = {**strip_not_given({"Prefer": prefer}), **(extra_headers or {})}
+        extra_headers = {
+            **strip_not_given(
+                {
+                    "Prefer": prefer,
+                    "Replicate-Max-Lifetime": replicate_max_lifetime,
+                }
+            ),
+            **(extra_headers or {}),
+        }
         return self._post(
             f"/models/{model_owner}/{model_name}/predictions",
             body=maybe_transform(
@@ -218,6 +227,7 @@ async def create(
         webhook: str | Omit = omit,
         webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | Omit = omit,
         prefer: str | Omit = omit,
+        replicate_max_lifetime: str | Omit = omit,
         file_encoding_strategy: Optional["FileEncodingStrategy"] = None,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
@@ -280,7 +290,7 @@ async def create(
               [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
               This field is no longer needed as the returned prediction will always have a
-              `stream` entry in its `url` property if the model supports streaming.
+              `stream` entry in its `urls` property if the model supports streaming.
 
           webhook: An HTTPS URL for receiving a webhook when the prediction has new output. The
               webhook will be a POST request where the request body is the same as the
@@ -331,7 +341,15 @@ async def create(
             raise ValueError(f"Expected a non-empty value for `model_owner` but received {model_owner!r}")
         if not model_name:
             raise ValueError(f"Expected a non-empty value for `model_name` but received {model_name!r}")
-        extra_headers = {**strip_not_given({"Prefer": prefer}), **(extra_headers or {})}
+        extra_headers = {
+            **strip_not_given(
+                {
+                    "Prefer": prefer,
+                    "Replicate-Max-Lifetime": replicate_max_lifetime,
+                }
+            ),
+            **(extra_headers or {}),
+        }
         return await self._post(
             f"/models/{model_owner}/{model_name}/predictions",
             body=await async_maybe_transform(
diff --git a/src/replicate/resources/predictions.py b/src/replicate/resources/predictions.py
index 04158d1..afd49a7 100644
--- a/src/replicate/resources/predictions.py
+++ b/src/replicate/resources/predictions.py
@@ -11,7 +11,7 @@
 from replicate.lib._files import FileEncodingStrategy, encode_json, async_encode_json
 
 from ..types import prediction_list_params, prediction_create_params
-from .._types import NOT_GIVEN, Body, Query, Headers, NotGiven
+from .._types import Body, Omit, Query, Headers, NotGiven, omit, not_given
 from .._utils import maybe_transform, strip_not_given, async_maybe_transform
 from .._compat import cached_property
 from .._resource import SyncAPIResource, AsyncAPIResource
@@ -63,17 +63,18 @@ def create(
         *,
         input: object,
         version: str,
-        stream: bool | NotGiven = NOT_GIVEN,
-        webhook: str | NotGiven = NOT_GIVEN,
-        webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | NotGiven = NOT_GIVEN,
-        prefer: str | NotGiven = NOT_GIVEN,
+        stream: bool | Omit = omit,
+        webhook: str | Omit = omit,
+        webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | Omit = omit,
+        prefer: str | Omit = omit,
+        replicate_max_lifetime: str | Omit = omit,
         file_encoding_strategy: Optional["FileEncodingStrategy"] = None,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> Prediction:
         """
         Create a prediction for the model version and inputs you provide.
@@ -138,7 +139,7 @@ def create(
               [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
               This field is no longer needed as the returned prediction will always have a
-              `stream` entry in its `url` property if the model supports streaming.
+              `stream` entry in its `urls` property if the model supports streaming.
 
           webhook: An HTTPS URL for receiving a webhook when the prediction has new output. The
               webhook will be a POST request where the request body is the same as the
@@ -186,7 +187,15 @@ def create(
 
           timeout: Override the client-level default timeout for this request, in seconds
         """
-        extra_headers = {**strip_not_given({"Prefer": prefer}), **(extra_headers or {})}
+        extra_headers = {
+            **strip_not_given(
+                {
+                    "Prefer": prefer,
+                    "Replicate-Max-Lifetime": replicate_max_lifetime,
+                }
+            ),
+            **(extra_headers or {}),
+        }
         return self._post(
             "/predictions",
             body=maybe_transform(
@@ -208,14 +217,14 @@ def create(
     def list(
         self,
         *,
-        created_after: Union[str, datetime] | NotGiven = NOT_GIVEN,
-        created_before: Union[str, datetime] | NotGiven = NOT_GIVEN,
+        created_after: Union[str, datetime] | Omit = omit,
+        created_before: Union[str, datetime] | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> SyncCursorURLPageWithCreatedFilters[Prediction]:
         """
         Get a paginated list of all predictions created by the user or organization
@@ -330,7 +339,7 @@ def cancel(
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> Prediction:
         """
         Cancel a prediction that is currently running.
@@ -386,7 +395,7 @@ def get(
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> Prediction:
         """
         Get the current state of a prediction.
@@ -448,7 +457,9 @@ def get(
         Terminated predictions (with a status of `succeeded`, `failed`, or `canceled`)
         will include a `metrics` object with a `predict_time` property showing the
         amount of CPU or GPU time, in seconds, that the prediction used while running.
-        It won't include time waiting for the prediction to start.
+        It won't include time waiting for the prediction to start. The `metrics` object
+        will also include a `total_time` property showing the total time, in seconds,
+        that the prediction took to complete.
 
         All input parameters, output values, and logs are automatically removed after an
         hour, by default, for predictions created through the API.
@@ -514,17 +525,18 @@ async def create(
         *,
         input: object,
         version: str,
-        stream: bool | NotGiven = NOT_GIVEN,
-        webhook: str | NotGiven = NOT_GIVEN,
-        webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | NotGiven = NOT_GIVEN,
-        prefer: str | NotGiven = NOT_GIVEN,
+        stream: bool | Omit = omit,
+        webhook: str | Omit = omit,
+        webhook_events_filter: List[Literal["start", "output", "logs", "completed"]] | Omit = omit,
+        prefer: str | Omit = omit,
+        replicate_max_lifetime: str | Omit = omit,
         file_encoding_strategy: Optional["FileEncodingStrategy"] = None,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> Prediction:
         """
         Create a prediction for the model version and inputs you provide.
@@ -589,7 +601,7 @@ async def create(
               [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
               This field is no longer needed as the returned prediction will always have a
-              `stream` entry in its `url` property if the model supports streaming.
+              `stream` entry in its `urls` property if the model supports streaming.
 
           webhook: An HTTPS URL for receiving a webhook when the prediction has new output. The
               webhook will be a POST request where the request body is the same as the
@@ -637,7 +649,15 @@ async def create(
 
           timeout: Override the client-level default timeout for this request, in seconds
         """
-        extra_headers = {**strip_not_given({"Prefer": prefer}), **(extra_headers or {})}
+        extra_headers = {
+            **strip_not_given(
+                {
+                    "Prefer": prefer,
+                    "Replicate-Max-Lifetime": replicate_max_lifetime,
+                }
+            ),
+            **(extra_headers or {}),
+        }
         return await self._post(
             "/predictions",
             body=await async_maybe_transform(
@@ -661,14 +681,14 @@ async def create(
     def list(
         self,
         *,
-        created_after: Union[str, datetime] | NotGiven = NOT_GIVEN,
-        created_before: Union[str, datetime] | NotGiven = NOT_GIVEN,
+        created_after: Union[str, datetime] | Omit = omit,
+        created_before: Union[str, datetime] | Omit = omit,
         # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
         # The extra values given here take precedence over values defined on the client or passed to this method.
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> AsyncPaginator[Prediction, AsyncCursorURLPageWithCreatedFilters[Prediction]]:
         """
         Get a paginated list of all predictions created by the user or organization
@@ -783,7 +803,7 @@ async def cancel(
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> Prediction:
         """
         Cancel a prediction that is currently running.
@@ -839,7 +859,7 @@ async def get(
         extra_headers: Headers | None = None,
         extra_query: Query | None = None,
         extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+        timeout: float | httpx.Timeout | None | NotGiven = not_given,
     ) -> Prediction:
         """
         Get the current state of a prediction.
@@ -901,7 +921,9 @@ async def get(
         Terminated predictions (with a status of `succeeded`, `failed`, or `canceled`)
         will include a `metrics` object with a `predict_time` property showing the
         amount of CPU or GPU time, in seconds, that the prediction used while running.
-        It won't include time waiting for the prediction to start.
+        It won't include time waiting for the prediction to start. The `metrics` object
+        will also include a `total_time` property showing the total time, in seconds,
+        that the prediction took to complete.
 
         All input parameters, output values, and logs are automatically removed after an
         hour, by default, for predictions created through the API.
diff --git a/src/replicate/resources/trainings.py b/src/replicate/resources/trainings.py
index 67600a9..a16b2f2 100644
--- a/src/replicate/resources/trainings.py
+++ b/src/replicate/resources/trainings.py
@@ -390,7 +390,9 @@ def get(
         Terminated trainings (with a status of `succeeded`, `failed`, or `canceled`)
         will include a `metrics` object with a `predict_time` property showing the
         amount of CPU or GPU time, in seconds, that the training used while running. It
-        won't include time waiting for the training to start.
+        won't include time waiting for the training to start. The `metrics` object will
+        also include a `total_time` property showing the total time, in seconds, that
+        the training took to complete.
 
         Args:
           extra_headers: Send extra headers
@@ -774,7 +776,9 @@ async def get(
         Terminated trainings (with a status of `succeeded`, `failed`, or `canceled`)
         will include a `metrics` object with a `predict_time` property showing the
         amount of CPU or GPU time, in seconds, that the training used while running. It
-        won't include time waiting for the training to start.
+        won't include time waiting for the training to start. The `metrics` object will
+        also include a `total_time` property showing the total time, in seconds, that
+        the training took to complete.
 
         Args:
           extra_headers: Send extra headers
diff --git a/src/replicate/types/collection_get_response.py b/src/replicate/types/collection_get_response.py
index 0573b66..396a839 100644
--- a/src/replicate/types/collection_get_response.py
+++ b/src/replicate/types/collection_get_response.py
@@ -67,3 +67,6 @@ class CollectionGetResponse(BaseModel):
 
     slug: str
     """The slug of the collection (lowercase with dashes)"""
+
+    full_description: Optional[str] = None
+    """The full description of the collection in markdown format"""
diff --git a/src/replicate/types/deployments/prediction_create_params.py b/src/replicate/types/deployments/prediction_create_params.py
index ed99336..6bc0863 100644
--- a/src/replicate/types/deployments/prediction_create_params.py
+++ b/src/replicate/types/deployments/prediction_create_params.py
@@ -47,7 +47,7 @@ class PredictionCreateParams(TypedDict, total=False):
     [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
     This field is no longer needed as the returned prediction will always have a
-    `stream` entry in its `url` property if the model supports streaming.
+    `stream` entry in its `urls` property if the model supports streaming.
     """
 
     webhook: str
@@ -94,3 +94,5 @@ class PredictionCreateParams(TypedDict, total=False):
     """
 
     prefer: Annotated[str, PropertyInfo(alias="Prefer")]
+
+    replicate_max_lifetime: Annotated[str, PropertyInfo(alias="Replicate-Max-Lifetime")]
diff --git a/src/replicate/types/models/prediction_create_params.py b/src/replicate/types/models/prediction_create_params.py
index 0427671..69f8d38 100644
--- a/src/replicate/types/models/prediction_create_params.py
+++ b/src/replicate/types/models/prediction_create_params.py
@@ -47,7 +47,7 @@ class PredictionCreateParams(TypedDict, total=False):
     [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
     This field is no longer needed as the returned prediction will always have a
-    `stream` entry in its `url` property if the model supports streaming.
+    `stream` entry in its `urls` property if the model supports streaming.
     """
 
     webhook: str
@@ -94,3 +94,5 @@ class PredictionCreateParams(TypedDict, total=False):
     """
 
     prefer: Annotated[str, PropertyInfo(alias="Prefer")]
+
+    replicate_max_lifetime: Annotated[str, PropertyInfo(alias="Replicate-Max-Lifetime")]
diff --git a/src/replicate/types/prediction.py b/src/replicate/types/prediction.py
index 4a5122b..83071e8 100644
--- a/src/replicate/types/prediction.py
+++ b/src/replicate/types/prediction.py
@@ -1,12 +1,14 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
-from typing import Dict, Union, Optional
+from typing import TYPE_CHECKING, Dict, Union, Optional
 from datetime import datetime
 from typing_extensions import Literal
 
+from pydantic import Field as FieldInfo
+
 from .._models import BaseModel
 
-__all__ = ["Prediction", "URLs"]
+__all__ = ["Prediction", "URLs", "Metrics"]
 
 
 class URLs(BaseModel):
@@ -23,6 +25,23 @@ class URLs(BaseModel):
     """An event source to stream the output of the prediction via API"""
 
 
+class Metrics(BaseModel):
+    total_time: Optional[float] = None
+    """The total time, in seconds, that the prediction took to complete"""
+
+    if TYPE_CHECKING:
+        # Some versions of Pydantic <2.8.0 have a bug and don’t allow assigning a
+        # value to this field, so for compatibility we avoid doing it at runtime.
+        __pydantic_extra__: Dict[str, object] = FieldInfo(init=False)  # pyright: ignore[reportIncompatibleVariableOverride]
+
+        # Stub to indicate that arbitrary properties are accepted.
+        # To access properties that are not valid identifiers you can use `getattr`, e.g.
+        # `getattr(obj, '$type')`
+        def __getattr__(self, attr: str) -> object: ...
+    else:
+        __pydantic_extra__: Dict[str, object]
+
+
 class Prediction(BaseModel):
     id: str
 
@@ -58,13 +77,19 @@ class Prediction(BaseModel):
     completed_at: Optional[datetime] = None
     """The time that the model completed the prediction and all outputs were uploaded"""
 
+    deadline: Optional[datetime] = None
+    """
+    The absolute time at which the prediction will be automatically canceled if it
+    has not completed
+    """
+
     deployment: Optional[str] = None
     """The name of the deployment that created the prediction"""
 
     logs: Optional[str] = None
     """The log output from the model"""
 
-    metrics: Optional[Dict[str, object]] = None
+    metrics: Optional[Metrics] = None
     """Additional metrics associated with the prediction"""
 
     started_at: Optional[datetime] = None
diff --git a/src/replicate/types/prediction_create_params.py b/src/replicate/types/prediction_create_params.py
index 8665e35..8578a34 100644
--- a/src/replicate/types/prediction_create_params.py
+++ b/src/replicate/types/prediction_create_params.py
@@ -43,7 +43,7 @@ class PredictionCreateParamsWithoutVersion(TypedDict, total=False):
     [server-sent events (SSE)](https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events).
 
     This field is no longer needed as the returned prediction will always have a
-    `stream` entry in its `url` property if the model supports streaming.
+    `stream` entry in its `urls` property if the model supports streaming.
     """
 
     webhook: str
@@ -92,6 +92,8 @@ class PredictionCreateParamsWithoutVersion(TypedDict, total=False):
 
     prefer: Annotated[str, PropertyInfo(alias="Prefer")]
 
+    replicate_max_lifetime: Annotated[str, PropertyInfo(alias="Replicate-Max-Lifetime")]
+
 
 class PredictionCreateParams(PredictionCreateParamsWithoutVersion):
     version: Required[str]
diff --git a/src/replicate/types/training_cancel_response.py b/src/replicate/types/training_cancel_response.py
index 9af512b..097d094 100644
--- a/src/replicate/types/training_cancel_response.py
+++ b/src/replicate/types/training_cancel_response.py
@@ -13,6 +13,9 @@ class Metrics(BaseModel):
     predict_time: Optional[float] = None
     """The amount of CPU or GPU time, in seconds, that the training used while running"""
 
+    total_time: Optional[float] = None
+    """The total time, in seconds, that the training took to complete"""
+
 
 class Output(BaseModel):
     version: Optional[str] = None
diff --git a/src/replicate/types/training_create_response.py b/src/replicate/types/training_create_response.py
index d29e3fa..02340dd 100644
--- a/src/replicate/types/training_create_response.py
+++ b/src/replicate/types/training_create_response.py
@@ -13,6 +13,9 @@ class Metrics(BaseModel):
     predict_time: Optional[float] = None
     """The amount of CPU or GPU time, in seconds, that the training used while running"""
 
+    total_time: Optional[float] = None
+    """The total time, in seconds, that the training took to complete"""
+
 
 class Output(BaseModel):
     version: Optional[str] = None
diff --git a/src/replicate/types/training_get_response.py b/src/replicate/types/training_get_response.py
index 4169da7..07b4571 100644
--- a/src/replicate/types/training_get_response.py
+++ b/src/replicate/types/training_get_response.py
@@ -13,6 +13,9 @@ class Metrics(BaseModel):
     predict_time: Optional[float] = None
     """The amount of CPU or GPU time, in seconds, that the training used while running"""
 
+    total_time: Optional[float] = None
+    """The total time, in seconds, that the training took to complete"""
+
 
 class Output(BaseModel):
     version: Optional[str] = None
diff --git a/src/replicate/types/training_list_response.py b/src/replicate/types/training_list_response.py
index 02adf3a..9313ede 100644
--- a/src/replicate/types/training_list_response.py
+++ b/src/replicate/types/training_list_response.py
@@ -13,6 +13,9 @@ class Metrics(BaseModel):
     predict_time: Optional[float] = None
     """The amount of CPU or GPU time, in seconds, that the training used while running"""
 
+    total_time: Optional[float] = None
+    """The total time, in seconds, that the training took to complete"""
+
 
 class Output(BaseModel):
     version: Optional[str] = None
diff --git a/tests/api_resources/deployments/test_predictions.py b/tests/api_resources/deployments/test_predictions.py
index c9d95e0..eeeed64 100644
--- a/tests/api_resources/deployments/test_predictions.py
+++ b/tests/api_resources/deployments/test_predictions.py
@@ -44,6 +44,7 @@ def test_method_create_with_all_params(self, client: Replicate) -> None:
             webhook="https://example.com/my-webhook-handler",
             webhook_events_filter=["start", "completed"],
             prefer="wait=5",
+            replicate_max_lifetime="5m",
         )
         assert_matches_type(Prediction, prediction, path=["response"])
 
@@ -139,6 +140,7 @@ async def test_method_create_with_all_params(self, async_client: AsyncReplicate)
             webhook="https://example.com/my-webhook-handler",
             webhook_events_filter=["start", "completed"],
             prefer="wait=5",
+            replicate_max_lifetime="5m",
         )
         assert_matches_type(Prediction, prediction, path=["response"])
 
diff --git a/tests/api_resources/models/test_predictions.py b/tests/api_resources/models/test_predictions.py
index 6c7f099..f702b05 100644
--- a/tests/api_resources/models/test_predictions.py
+++ b/tests/api_resources/models/test_predictions.py
@@ -44,6 +44,7 @@ def test_method_create_with_all_params(self, client: Replicate) -> None:
             webhook="https://example.com/my-webhook-handler",
             webhook_events_filter=["start", "completed"],
             prefer="wait=5",
+            replicate_max_lifetime="5m",
         )
         assert_matches_type(Prediction, prediction, path=["response"])
 
@@ -139,6 +140,7 @@ async def test_method_create_with_all_params(self, async_client: AsyncReplicate)
             webhook="https://example.com/my-webhook-handler",
             webhook_events_filter=["start", "completed"],
             prefer="wait=5",
+            replicate_max_lifetime="5m",
         )
         assert_matches_type(Prediction, prediction, path=["response"])
 
diff --git a/tests/api_resources/test_predictions.py b/tests/api_resources/test_predictions.py
index 1f0498a..e1640b2 100644
--- a/tests/api_resources/test_predictions.py
+++ b/tests/api_resources/test_predictions.py
@@ -38,6 +38,7 @@ def test_method_create_with_all_params(self, client: Replicate) -> None:
             webhook="https://example.com/my-webhook-handler",
             webhook_events_filter=["start", "completed"],
             prefer="wait=5",
+            replicate_max_lifetime="5m",
         )
         assert_matches_type(Prediction, prediction, path=["response"])
 
@@ -215,6 +216,7 @@ async def test_method_create_with_all_params(self, async_client: AsyncReplicate)
             webhook="https://example.com/my-webhook-handler",
             webhook_events_filter=["start", "completed"],
             prefer="wait=5",
+            replicate_max_lifetime="5m",
         )
         assert_matches_type(Prediction, prediction, path=["response"])