alexanto · Nov 5, 2026 · Nov 5, 2026 · Nov 6, 2026
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,8 +1,9 @@
 ## Unreleased (v7.0.0)
 * (Major): Python 3.10 or newer is required. Support for Python 3.8 and 3.9 has been dropped.
-* (Major): `UnleashClient` and `AsyncUnleashClient` take their settings as option groups: `identity=IdentityOptions(...)`, `http=HttpOptions(...)`, `mode=PollingOptions(...)` or `mode=StreamingOptions()`, and `metrics=MetricsOptions(...)`. Every argument after `url` and `app_name` is keyword-only. `ExperimentalMode` has been removed. See the [v7 migration guide](./v7_MIGRATION_GUIDE.md).
+* (Major): `UnleashClient` and `AsyncUnleashClient` take their settings as option groups: `identity=IdentityOptions(...)`, `mode=PollingOptions(...)` or `mode=StreamingOptions()`, and `metrics=MetricsOptions(...)`. Every argument after `url` and `app_name` is keyword-only. `ExperimentalMode` has been removed. See the [v7 migration guide](./v7_MIGRATION_GUIDE.md).
 * (Major): Settings can no longer be changed on an existing client. The `unleash_*` settings properties and the `strategy_mapping` and `connector_mode` properties have been removed, together with their setters.
-* (Major): The API token is passed as `api_key="<API token>"`. An `Authorization` header in `HttpOptions.headers` raises `ValueError`.
+* (Major): The API token is passed as `api_key="<API token>"`. An `Authorization` header in a transport's `headers` raises `ValueError`.
+* (Major): HTTP settings are set on the transport: `custom_headers` and `request_timeout` become `RequestsTransport(headers=..., timeout=...)`, or `AiohttpTransport(...)` for `AsyncUnleashClient`. `custom_options` has been removed: configure a `requests.Session` and pass `RequestsTransport(session)`. `request_retries` has been removed: a failed feature fetch is retried up to 3 times. The streaming connection doesn't use the transport: it is sent with the SDK's own headers and a 30s timeout.
 * (Major): The deprecated `environment` setting has been removed. The `environment` field of the evaluation context and the impact metrics `environment` label come from the API token, or are `"default"` when the token doesn't name one.
 * (Major): The `UnleashClient` and `AsyncUnleashClient` constructors raise `ValueError` if `url` is not an absolute http(s) URL with a host, for every client.
 * (Major): The constructors raise `ValueError` if `api_key`, `app_name` or `IdentityOptions.instance_id` contains a line break, such as an API token read from a file with a trailing newline, and `TypeError` if one isn't a str. These values are sent as headers.

diff --git a/README.md b/README.md
@@ -117,14 +117,14 @@ The client will evaluate the fallback function if the feature flag is not found
 
 ```python
 from UnleashClient import (
-    HttpOptions, MetricsOptions, PollingOptions, UnleashClient,
+    MetricsOptions, PollingOptions, RequestsTransport, UnleashClient,
 )
 
 client = UnleashClient(
     url="https://YOUR-API-URL",
     app_name="my-python-app",
     api_key="<API token>",
-    http=HttpOptions(timeout=10),
+    transport=RequestsTransport(timeout=10),
     mode=PollingOptions(interval=30),
     metrics=MetricsOptions(disabled=True),
 )
@@ -138,7 +138,6 @@ Settings can't be changed after the client is created.
 | app_name               | Name of the application using the client. Required.                                                                                       | —                   |
 | api_key                | Client API token, sent as the `Authorization` header. | None                |
 | identity               | How the client identifies itself. See [`IdentityOptions`](#identityoptions).                                                              | `IdentityOptions()` |
-| http                   | How the client talks to the server. See [`HttpOptions`](#httpoptions).                                                                    | `HttpOptions()`     |
 | mode                   | How the client receives flag updates: [`PollingOptions`](#pollingoptions) or [`StreamingOptions`](#streamingoptions).                     | `PollingOptions()`  |
 | metrics                | How the client reports to the server. See [`MetricsOptions`](#metricsoptions).                                                            | `MetricsOptions()`  |
 | verbose_log_level      | [Log level](https://docs.python.org/3/library/logging.html#logging-levels) for messages about unknown flags and failed impression events. | 30                  |
@@ -149,7 +148,7 @@ Settings can't be changed after the client is created.
 | scheduler_executor     | Name of the executor to use on your custom `scheduler`. Required with `scheduler`. Sync client only.                                      | None                |
 | multiple_instance_mode | How to handle multiple client instances (`BLOCK`, `WARN`, `SILENTLY_ALLOW`).                                                              | `WARN`              |
 | event_callback         | Function that receives impression and lifecycle events. See [events and impression data](#events-and-impression-data).                    | None                |
-| transport              | How requests are sent. Can't be combined with `http` (which will be removed). See [custom HTTP transport](#custom-http-transport).    | `RequestsTransport` or `AiohttpTransport` |
+| transport              | How requests are sent, including headers, timeout, proxy and TLS settings. See [custom HTTP transport](#custom-http-transport).    | `RequestsTransport` or `AiohttpTransport` |
 
 #### `IdentityOptions`
 
@@ -159,15 +158,6 @@ Settings can't be changed after the client is created.
 | sdk_flavor         | Identifier of an integration built on this SDK (e.g. an OpenFeature provider).         | None                 |
 | sdk_flavor_version | Version of the integration named by `sdk_flavor`.                                      | None                 |
 
-#### `HttpOptions`
-
-| Parameter | Description                                                                                                                        | Default |
-| --------- | ---------------------------------------------------------------------------------------------------------------------------------- | ------- |
-| headers   | Extra HTTP headers sent with every request. Pass the API token as `api_key`; an `Authorization` header here raises `ValueError`.   | None    |
-| options   | Extra keyword arguments for every [HTTP request](https://requests.readthedocs.io/en/latest/api/#main-interface). Sync client only. | None    |
-| timeout   | HTTP request timeout (seconds).                                                                                                    | 30      |
-| retries   | How many times a failed feature fetch is retried.                                                                                  | 3       |
-
 #### `PollingOptions`
 
 The default: the client fetches feature flags from the server on an interval.
@@ -531,7 +521,7 @@ tls.load_cert_chain("/etc/unleash/client.crt", "/etc/unleash/client.key")
 
 async def main():
     async with aiohttp.ClientSession(
-        trust_env=True,  
+        trust_env=True,
         connector=aiohttp.TCPConnector(ssl=tls),
         trace_configs=[my_trace_config],  # e.g. aiohttp tracing for OpenTelemetry
     ) as session:
@@ -671,11 +661,11 @@ All four examples are in [`examples/`](examples/) and run as part of the SDK's t
 
 #### Things to know
 
-- **Don't retry in your transport.** The client already retries failed feature fetches, with a backoff. A `requests` session with a retrying adapter mounted logs a warning, because failed fetches would be retried twice over. aiohttp has no retry setting, but resends a GET once by itself when the server drops a reused connection; that's expected, and not a retry to turn off.
+- **Don't retry in your transport.** The client already retries a failed feature fetch up to 3 times, with a backoff. A `requests` session with a retrying adapter mounted logs a warning, because failed fetches would be retried twice over. aiohttp has no retry setting, but resends a GET once by itself when the server drops a reused connection; that's expected, and not a retry to turn off.
 - **Don't log request headers.** They include your API token.
 - **Don't add an HTTP cache.** The client already sends `If-None-Match` and handles `304 Not Modified`. A caching session (such as CacheControl or requests-cache) turns each 304 into its cached 200, so the client reloads unchanged flags on every poll. The flags stay correct.
 - **Create the session in the process that uses it.** With a forking server, such as gunicorn with `--preload` or uWSGI without `lazy-apps`, build the session together with the client in each worker. Forked processes must not share open connections. See [running in multi-process setups](#running-in-multi-process-setups).
-- **Streaming mode doesn't use the transport.** With `mode=StreamingOptions()`, the stream connects with the client's default HTTP settings.
+- **Streaming mode doesn't use the transport.** With `mode=StreamingOptions()`, the stream connects with the SDK's own headers and a 30 second timeout. Custom transport's headers, timeout, proxy and TLS settings don't apply to it.
 
 ## Running in multi-process setups
 

diff --git a/UnleashClient/__init__.py b/UnleashClient/__init__.py
@@ -7,7 +7,6 @@
     build_ready_callback,
 )
 from .config import (
-    HttpOptions,
     IdentityOptions,
     MetricsOptions,
     PollingOptions,

diff --git a/UnleashClient/_async_unleash_api_client.py b/UnleashClient/_async_unleash_api_client.py
@@ -38,23 +38,21 @@ def __init__(
         transport: Optional[AsyncTransport] = None,
     ) -> None:
         """
-        :param config: read for the url, timeouts, retries and project.
+        :param config: read for the url and the identity.
         :param headers: builds the header set each request needs.
         :param events: receives ``FETCH_COMPLETED`` after every feature fetch.
         :param transport: sends the requests. Defaults to an
-                          :class:`AiohttpTransport` built from ``config.http``,
-                          which :meth:`aclose` closes. A transport passed in is
-                          left open: whoever created it closes it.
+                          :class:`AiohttpTransport`, which :meth:`aclose`
+                          closes. A transport passed in is left open: whoever
+                          created it closes it.
         """
         self._config: UnleashConfig = config
         self._headers: _HeaderFactory = headers
         self._events: Optional[_EventDispatcher] = events
         # set only when this API client built the transport, and so it closes it.
         self._default_transport: Optional[AiohttpTransport] = None
         if transport is None:
-            transport = self._default_transport = AiohttpTransport(
-                timeout=config.http.timeout
-            )
+            transport = self._default_transport = AiohttpTransport()
         self._transport: AsyncTransport = transport
         self._is_closed: bool = False
         self._close_lock = threading.Lock()
@@ -128,7 +126,6 @@ async def _fetch_features(
                 Request(
                     method="GET", url=base_url, headers=headers, params=base_params
                 ),
-                config.http.retries,
             )
 
             http_status_code = resp.status

diff --git a/UnleashClient/_headers.py b/UnleashClient/_headers.py
@@ -24,11 +24,6 @@ def __init__(self, config: UnleashConfig) -> None:
 
     def base(self) -> Dict[str, str]:
         return {
-            # Custom headers go first: the application and identification
-            # headers below are not overridable.
-            **self._config.http.headers,
-            # HttpOptions rejects an Authorization header, so the custom
-            # headers cannot override the API key.
             **({"Authorization": self._config.api_key} if self._config.api_key else {}),
             **APPLICATION_HEADERS,
             "unleash-connection-id": self._config.connection_id,

diff --git a/UnleashClient/_retry.py b/UnleashClient/_retry.py
@@ -8,13 +8,14 @@
 import asyncio
 import random
 import time
-from typing import Union
+from typing import Optional, Union
 
 from UnleashClient.errors import AlreadyClosedError
 from UnleashClient.transport import AsyncTransport, Request, Response, Transport
 from UnleashClient.utils import LOGGER
 
 RETRY_STATUSES = frozenset({500, 502, 504})
+FETCH_RETRIES = 3
 BACKOFF_BASE = 0.5
 
 _sleep = time.sleep
@@ -38,11 +39,11 @@ def _backoff_delay(retry: int) -> float:
 
 def _attempts(retries: int) -> range:
     """The attempt numbers: the first attempt and ``retries`` retries."""
-    return range(1 + max(retries, 0))
+    return range(1 + retries)
 
 
 def _should_retry(attempt: int, retries: int, outcome: _Outcome) -> bool:
-    if attempt >= max(retries, 0):
+    if attempt >= retries:
         return False
     if isinstance(outcome, Response):
         return outcome.status in RETRY_STATUSES
@@ -66,16 +67,18 @@ def _next_step(attempt: int, retries: int, outcome: _Outcome) -> Union[Response,
 
 
 def _send_with_retries(
-    transport: Transport, request: Request, retries: int
+    transport: Transport, request: Request, retries: Optional[int] = None
 ) -> Response:
     """
     Send ``request``, retrying on a status in :data:`RETRY_STATUSES` and on any
     exception, with a backoff between attempts.ca
 
-    :param retries: Retries after the first attempt; negative counts as 0.
+    :param retries: Retries after the first attempt; :data:`FETCH_RETRIES` when
+                    None
     :return: The last response, which can still have a retry status.
     :raises Exception: The last attempt's exception, if it raised.
     """
+    retries = FETCH_RETRIES if retries is None else retries
     for attempt in _attempts(retries):
         try:
             outcome: _Outcome = transport.execute(request)
@@ -90,9 +93,10 @@ def _send_with_retries(
 
 
 async def _send_with_retries_async(
-    transport: AsyncTransport, request: Request, retries: int
+    transport: AsyncTransport, request: Request, retries: Optional[int] = None
 ) -> Response:
     """The asyncio twin of :func:`_send_with_retries`: only the awaits differ."""
+    retries = FETCH_RETRIES if retries is None else retries
     for attempt in _attempts(retries):
         try:
             outcome: _Outcome = await transport.execute(request)

diff --git a/UnleashClient/_unleash_api_client.py b/UnleashClient/_unleash_api_client.py
@@ -101,8 +101,7 @@ class _UnleashApiClient:
     :param config: Client configuration.
     :param headers: Builds the header set each endpoint needs.
     :param events: Receives ``FETCH_COMPLETED`` after every feature fetch.
-    :param transport: Sends the requests. Defaults to a :class:`RequestsTransport`
-                      built from ``config.http``. A transport passed in is never
+    :param transport: Sends the requests. Defaults to a :class:`RequestsTransport`. A transport passed in is never
                       closed: whoever created it closes it.
     """
 
@@ -117,11 +116,7 @@ def __init__(
         self._headers: _HeaderFactory = headers
         self._events: Optional[_EventDispatcher] = events
 
-        self._transport: Transport = transport or RequestsTransport(
-            timeout=config.http.timeout,
-            # this is going to go away soon, just like the whole httpOptions in the client constructor
-            _request_kwargs=config.http.options,
-        )
+        self._transport: Transport = transport or RequestsTransport()
 
     # pylint: disable=broad-except
     def fetch_features(self, etag: str = "", *, options: PollingOptions) -> FetchResult:
@@ -177,7 +172,6 @@ def _fetch_features(
                     },
                     params=base_params,
                 ),
-                config.http.retries,
             )
 
             http_status_code = resp.status

diff --git a/UnleashClient/clients/async_unleash_client.py b/UnleashClient/clients/async_unleash_client.py
@@ -27,7 +27,6 @@
 from UnleashClient.cache import BaseCache, FileCache
 from UnleashClient.clients.unleash_client import _RunState
 from UnleashClient.config import (
-    HttpOptions,
     IdentityOptions,
     MetricsOptions,
     PollingOptions,
@@ -71,7 +70,6 @@ class AsyncUnleashClient:
     :param app_name: Name of the application using the client, required.
     :param api_key: Client API token.  Optional, because a client that only loads from a backup never talks to the server.
     :param identity: How the client identifies itself.
-    :param http: How the client talks to the server.
     :param mode: How often the client polls for flag updates.  Streaming is not supported yet.
     :param metrics: How the client reports to the server.
     :param verbose_log_level: Numerical log level (https://docs.python.org/3/library/logging.html#logging-levels) for cases where checking a feature flag fails.
@@ -80,7 +78,7 @@ class AsyncUnleashClient:
     :param cache_directory: Location of the cache directory.
     :param multiple_instance_mode: Determines how multiple instances being instantiated is handled by the SDK.  Defaults to InstanceAllowType.WARN.
     :param event_callback: Function to call if impression events are enabled.  Called on a dedicated background thread.
-    :param transport: Sends the requests to the server, see :class:`~UnleashClient.transport.AsyncTransport`.  Pass an :class:`AiohttpTransport` with your own ``aiohttp.ClientSession`` for a proxy, TLS settings or tracing.  The client never closes it.  Can't be combined with ``http``.
+    :param transport: Sends the requests to the server, see :class:`~UnleashClient.transport.AsyncTransport`.  Pass an :class:`AiohttpTransport` with your own ``aiohttp.ClientSession`` for a proxy, TLS settings or tracing.  The client never closes it.
     """
 
     def __init__(
@@ -90,7 +88,6 @@ def __init__(
         *,
         api_key: Optional[str] = None,
         identity: Optional[IdentityOptions] = None,
-        http: Optional[HttpOptions] = None,
         mode: Optional[PollingOptions] = None,
         metrics: Optional[MetricsOptions] = None,
         verbose_log_level: int = 30,
@@ -104,7 +101,6 @@ def __init__(
         if transport is not None:
             _check_transport(
                 transport,
-                http_given=http is not None,
                 is_async=True,
                 session_type=aiohttp.ClientSession,
                 wrapper_name="AiohttpTransport",
@@ -119,7 +115,6 @@ def __init__(
             app_name=app_name,
             api_key=api_key,
             identity=identity,
-            http=http,
             mode=mode,
             metrics=metrics,
             verbose_log_level=verbose_log_level,