diff --git a/litellm-rust/Cargo.lock b/litellm-rust/Cargo.lock
index 83cdbc6a782..5fbddcaffcf 100644
--- a/litellm-rust/Cargo.lock
+++ b/litellm-rust/Cargo.lock
@@ -2141,6 +2141,7 @@ dependencies = [
  "reqwest 0.12.28",
  "rstest",
  "rustls 0.23.42",
+ "serde_json",
  "thiserror 2.0.19",
  "tokio",
  "webpki-roots",
diff --git a/litellm-rust/crates/core/AGENTS.md b/litellm-rust/crates/core/AGENTS.md
index 449c3e647f7..0c8a747019d 100644
--- a/litellm-rust/crates/core/AGENTS.md
+++ b/litellm-rust/crates/core/AGENTS.md
@@ -5,10 +5,11 @@ litellm-core is the LiteLLM SDK in Rust — it makes the LLM call. Each top-leve
 Each crate mirrors one top-level Python package, so a Rust path reads as its Python path with the crate name in place of the package directory. Dependencies only point down:
 
 - `litellm-types` mirrors `litellm/types/`: pure serde data, no I/O
-- `litellm-core-utils` mirrors `litellm/litellm_core_utils/`: pure helpers (provider resolution, prompt factory, call arguments), no network I/O
-- `litellm-llms` mirrors `litellm/llms/`: `base_llm/<api>/transformation.rs`, `<provider>/<api>/transformation.rs`, and `custom_httpx/` (HTTP helpers and the OCR request handler)
+- `litellm-core-utils` mirrors `litellm/litellm_core_utils/`: pure helpers (provider resolution, prompt factory, call arguments, settings lookup and layer merge), no network I/O
+- `litellm-http` is Rust-only and route-neutral: settings resolution, the pooled `reqwest` clients, TLS, proxies, the SSRF-safe media fetcher, request and header helpers, and transport errors. Python's `litellm/llms/custom_httpx/` is split by responsibility instead of mirrored: its transport half lives here, its OCR handler in `litellm-llms`
+- `litellm-llms` mirrors `litellm/llms/`: `base_llm/<api>/transformation.rs`, `<provider>/<api>/transformation.rs`, and `base_llm/ocr/handler.rs` (the OCR request handler)
 - `litellm-core` mirrors the route packages (`litellm/ocr/`, `litellm/messages/`, ...): entrypoints, route request types, provider dispatch, the route machine, and hooks
 
-A route module owns the call entrypoint, route request types (`*Request<'a>`), credential fallback, provider dispatch, and the handler glue that runs a provider config. Provider code never imports from core; when it needs the caller's hooks mid-call it goes through `litellm_llms::custom_httpx::llm_http_handler::CallHooks`, which each route implements over its host. Import every item from its canonical path. Never re-export another crate's items or give an item a second public path; the only re-export allowed is a private submodule surfacing its item at its module root (`mod error; pub use error::Error;`). Handlers belong in core or llms, never in a host crate
+A route module owns the call entrypoint, route request types (`*Request<'a>`), credential fallback, provider dispatch, and the handler glue that runs a provider config. Provider code never imports from core; when it needs the caller's hooks mid-call it goes through `litellm_llms::base_llm::ocr::handler::CallHooks`, which each route implements over its host. Import every item from its canonical path. Never re-export another crate's items or give an item a second public path; the only re-export allowed is a private submodule surfacing its item at its module root (`mod error; pub use error::Error;`). Handlers belong in core or llms, never in a host crate
 
 Not here: serving HTTP (axum routes, extractors), config file reading, rollout state, databases, or callback execution of any kind. Core runs each route as a machine that yields host operations and call events; which integrations consume those events is the host's business.
diff --git a/litellm-rust/crates/core/Cargo.toml b/litellm-rust/crates/core/Cargo.toml
index ab04fb8d4ae..69ae8004d46 100644
--- a/litellm-rust/crates/core/Cargo.toml
+++ b/litellm-rust/crates/core/Cargo.toml
@@ -15,6 +15,7 @@ futures-util.workspace = true
 base64.workspace = true
 litellm-auth.workspace = true
 litellm-auth-aws.workspace = true
+litellm-http.workspace = true
 litellm-llms.workspace = true
 moka.workspace = true
 mime_guess = "2.0.5"
@@ -36,7 +37,6 @@ veil.workspace = true
 
 [dev-dependencies]
 litellm-auth-gcp.workspace = true
-litellm-http.workspace = true
 litellm-llms = { workspace = true, features = ["test-support"] }
 rstest.workspace = true
 rstest_reuse.workspace = true
diff --git a/litellm-rust/crates/core/src/audio_transcription/error.rs b/litellm-rust/crates/core/src/audio_transcription/error.rs
index 39b08e882f5..122cbab358f 100644
--- a/litellm-rust/crates/core/src/audio_transcription/error.rs
+++ b/litellm-rust/crates/core/src/audio_transcription/error.rs
@@ -20,9 +20,9 @@ pub enum Error {
     #[error(transparent)]
     Auth(#[from] litellm_auth::Error),
     #[error(transparent)]
-    Transport(#[from] litellm_llms::custom_httpx::transport::Error),
+    Transport(#[from] litellm_http::transport::Error),
     #[error(transparent)]
-    Headers(#[from] litellm_llms::custom_httpx::http_handler::HeaderError),
+    Headers(#[from] litellm_http::request::HeaderError),
     #[error(transparent)]
     Aws(#[from] litellm_auth_aws::Error),
 }
diff --git a/litellm-rust/crates/core/src/audio_transcription/handler.rs b/litellm-rust/crates/core/src/audio_transcription/handler.rs
index 0704f9391b0..503cc922966 100644
--- a/litellm-rust/crates/core/src/audio_transcription/handler.rs
+++ b/litellm-rust/crates/core/src/audio_transcription/handler.rs
@@ -1,4 +1,4 @@
-use litellm_llms::custom_httpx::http_handler::{http_request, truncate_error_body};
+use litellm_http::request::{http_request, truncate_error_body};
 use serde_json::Value;
 
 use super::{Error, client::http_client};
@@ -18,23 +18,17 @@ pub async fn execute_audio_transcription_provider_call(
         request_builder = request_builder.timeout(duration);
     }
     let response = http_request(request_builder).await.map_err(|error| {
-        Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-            error.to_string(),
-        ))
+        Error::Transport(litellm_http::transport::Error::Network(error.to_string()))
     })?;
     let status = response.status();
     let text = response.text().await.map_err(|error| {
-        Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-            error.to_string(),
-        ))
+        Error::Transport(litellm_http::transport::Error::Network(error.to_string()))
     })?;
     if !status.is_success() {
-        return Err(Error::Transport(
-            litellm_llms::custom_httpx::transport::Error::Http {
-                status: status.as_u16(),
-                body: truncate_error_body(&text),
-            },
-        ));
+        return Err(Error::Transport(litellm_http::transport::Error::Http {
+            status: status.as_u16(),
+            body: truncate_error_body(&text),
+        }));
     }
     let response_json = serde_json::from_str(&text)
         .map_err(|error| Error::InvalidResponse(format!("invalid audio response JSON: {error}")))?;
diff --git a/litellm-rust/crates/core/src/audio_transcription/prepare.rs b/litellm-rust/crates/core/src/audio_transcription/prepare.rs
index 193122db733..829617d26bd 100644
--- a/litellm-rust/crates/core/src/audio_transcription/prepare.rs
+++ b/litellm-rust/crates/core/src/audio_transcription/prepare.rs
@@ -1,10 +1,10 @@
 use litellm_core_utils::get_llm_provider_logic::{CustomLlmProvider, get_custom_llm_provider};
+use litellm_http::request::{has_header, string_headers};
 use litellm_llms::{
     base_llm::audio_transcription::transformation::{
         AudioTranscriptionAuth, BaseAudioTranscriptionConfig,
     },
     bedrock::audio_transcription::BEDROCK_AUDIO_TRANSCRIPTION_CONFIG,
-    custom_httpx::http_handler::{has_header, string_headers},
 };
 
 use super::Error;
diff --git a/litellm-rust/crates/core/src/chat_completions/common_utils.rs b/litellm-rust/crates/core/src/chat_completions/common_utils.rs
index cc9459793df..4ed39a90366 100644
--- a/litellm-rust/crates/core/src/chat_completions/common_utils.rs
+++ b/litellm-rust/crates/core/src/chat_completions/common_utils.rs
@@ -1,8 +1,8 @@
+use litellm_http::request::string_headers as shared_string_headers;
 use litellm_llms::{
     anthropic::chat::transformation::ANTHROPIC_CHAT_COMPLETIONS_CONFIG,
     base_llm::chat::transformation::BaseConfig,
     bedrock::chat::converse_transformation::BEDROCK_CHAT_COMPLETIONS_CONFIG,
-    custom_httpx::http_handler::string_headers as shared_string_headers,
 };
 use serde_json::{Map, Value};
 
diff --git a/litellm-rust/crates/core/src/chat_completions/error.rs b/litellm-rust/crates/core/src/chat_completions/error.rs
index 39b08e882f5..122cbab358f 100644
--- a/litellm-rust/crates/core/src/chat_completions/error.rs
+++ b/litellm-rust/crates/core/src/chat_completions/error.rs
@@ -20,9 +20,9 @@ pub enum Error {
     #[error(transparent)]
     Auth(#[from] litellm_auth::Error),
     #[error(transparent)]
-    Transport(#[from] litellm_llms::custom_httpx::transport::Error),
+    Transport(#[from] litellm_http::transport::Error),
     #[error(transparent)]
-    Headers(#[from] litellm_llms::custom_httpx::http_handler::HeaderError),
+    Headers(#[from] litellm_http::request::HeaderError),
     #[error(transparent)]
     Aws(#[from] litellm_auth_aws::Error),
 }
diff --git a/litellm-rust/crates/core/src/chat_completions/handler.rs b/litellm-rust/crates/core/src/chat_completions/handler.rs
index 034408bdf17..b73d4838760 100644
--- a/litellm-rust/crates/core/src/chat_completions/handler.rs
+++ b/litellm-rust/crates/core/src/chat_completions/handler.rs
@@ -1,7 +1,5 @@
-use litellm_llms::{
-    base_llm::chat::transformation::{ChatCompletionsAuth, ProviderChatResponseData},
-    custom_httpx::http_handler::{http_request, truncate_error_body},
-};
+use litellm_http::request::{http_request, truncate_error_body};
+use litellm_llms::base_llm::chat::transformation::{ChatCompletionsAuth, ProviderChatResponseData};
 use litellm_types::utils::ChatCompletionsResponse;
 use serde_json::Value;
 
@@ -34,30 +32,22 @@ pub(super) async fn execute_chat_completions_provider_call(
         // so the host can still serve it. Everything else here, a timeout
         // above all, may have reached the provider and been answered.
         if err.is_connect() || err.is_builder() {
-            Error::Transport(litellm_llms::custom_httpx::transport::Error::Connect(
-                err.to_string(),
-            ))
+            Error::Transport(litellm_http::transport::Error::Connect(err.to_string()))
         } else {
-            Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-                err.to_string(),
-            ))
+            Error::Transport(litellm_http::transport::Error::Network(err.to_string()))
         }
     })?;
 
     let status = response.status();
     let text = response.text().await.map_err(|err| {
-        Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-            err.to_string(),
-        ))
+        Error::Transport(litellm_http::transport::Error::Network(err.to_string()))
     })?;
 
     if !status.is_success() {
-        return Err(Error::Transport(
-            litellm_llms::custom_httpx::transport::Error::Http {
-                status: status.as_u16(),
-                body: truncate_error_body(&text),
-            },
-        ));
+        return Err(Error::Transport(litellm_http::transport::Error::Http {
+            status: status.as_u16(),
+            body: truncate_error_body(&text),
+        }));
     }
 
     let body: Value = serde_json::from_str(&text).map_err(|err| {
@@ -82,9 +72,7 @@ pub(super) async fn execute_chat_completions_provider_call(
 pub(super) fn as_response_error(err: Error) -> Error {
     match err {
         already @ (Error::InvalidResponse(_)
-        | Error::Transport(litellm_llms::custom_httpx::transport::Error::Http {
-            ..
-        })) => already,
+        | Error::Transport(litellm_http::transport::Error::Http { .. })) => already,
         other => Error::InvalidResponse(other.to_string()),
     }
 }
diff --git a/litellm-rust/crates/core/src/chat_completions/prepare.rs b/litellm-rust/crates/core/src/chat_completions/prepare.rs
index d408ea6574e..d0aa1e88011 100644
--- a/litellm-rust/crates/core/src/chat_completions/prepare.rs
+++ b/litellm-rust/crates/core/src/chat_completions/prepare.rs
@@ -1,8 +1,6 @@
 use litellm_core_utils::get_llm_provider_logic::{CustomLlmProvider, get_custom_llm_provider};
-use litellm_llms::{
-    base_llm::chat::transformation::{BaseConfig, ChatCompletionsAuth},
-    custom_httpx::http_handler::has_header,
-};
+use litellm_http::request::has_header;
+use litellm_llms::base_llm::chat::transformation::{BaseConfig, ChatCompletionsAuth};
 use litellm_types::llms::openai::ChatMessage;
 use serde_json::Value;
 
diff --git a/litellm-rust/crates/core/src/chat_completions/tests.rs b/litellm-rust/crates/core/src/chat_completions/tests.rs
index cbc4995ce0d..dcaa3397add 100644
--- a/litellm-rust/crates/core/src/chat_completions/tests.rs
+++ b/litellm-rust/crates/core/src/chat_completions/tests.rs
@@ -265,7 +265,7 @@ fn rejects_non_string_extra_headers() {
     call.extra_headers = Some(Map::from_iter([("x-trace".to_string(), json!(7))]));
     assert_eq!(
         decline(call),
-        Error::Headers(litellm_llms::custom_httpx::http_handler::HeaderError {
+        Error::Headers(litellm_http::request::HeaderError {
             context: "chat completions",
             name: "x-trace".to_string(),
             actual: "number",
@@ -771,10 +771,7 @@ mod round_trip {
         assert!(
             matches!(
                 err,
-                Error::Transport(litellm_llms::custom_httpx::transport::Error::Http {
-                    status: 429,
-                    ..
-                })
+                Error::Transport(litellm_http::transport::Error::Http { status: 429, .. })
             ),
             "expected a 429, got {err:?}"
         );
@@ -801,7 +798,7 @@ mod round_trip {
         assert!(
             matches!(
                 err,
-                Error::Transport(litellm_llms::custom_httpx::transport::Error::Connect(_))
+                Error::Transport(litellm_http::transport::Error::Connect(_))
             ),
             "expected a pre-send connect failure, got {err:?}"
         );
@@ -825,16 +822,11 @@ mod round_trip {
         }
         // An upstream status is already unambiguous, so it survives intact.
         assert!(matches!(
-            as_response_error(Error::Transport(
-                litellm_llms::custom_httpx::transport::Error::Http {
-                    status: 500,
-                    body: "boom".to_string()
-                }
-            )),
-            Error::Transport(litellm_llms::custom_httpx::transport::Error::Http {
+            as_response_error(Error::Transport(litellm_http::transport::Error::Http {
                 status: 500,
-                ..
-            })
+                body: "boom".to_string()
+            })),
+            Error::Transport(litellm_http::transport::Error::Http { status: 500, .. })
         ));
     }
 }
diff --git a/litellm-rust/crates/core/src/messages/common_utils.rs b/litellm-rust/crates/core/src/messages/common_utils.rs
index ec392324784..dcefa3ebffc 100644
--- a/litellm-rust/crates/core/src/messages/common_utils.rs
+++ b/litellm-rust/crates/core/src/messages/common_utils.rs
@@ -1,11 +1,9 @@
-pub(super) use litellm_llms::custom_httpx::http_handler::{
-    has_bearer_auth, has_header, truncate_error_body,
-};
+use litellm_http::request::string_headers as shared_string_headers;
+pub(super) use litellm_http::request::{has_bearer_auth, has_header, truncate_error_body};
 use litellm_llms::{
     anthropic::experimental_pass_through::messages::transformation::ANTHROPIC_MESSAGES_CONFIG,
     azure_ai::anthropic::messages_transformation::AZURE_ANTHROPIC_MESSAGES_CONFIG,
     base_llm::anthropic_messages::transformation::BaseAnthropicMessagesConfig,
-    custom_httpx::http_handler::string_headers as shared_string_headers,
 };
 use serde_json::{Map, Value};
 
diff --git a/litellm-rust/crates/core/src/messages/error.rs b/litellm-rust/crates/core/src/messages/error.rs
index 71bb748c50d..51fb764032c 100644
--- a/litellm-rust/crates/core/src/messages/error.rs
+++ b/litellm-rust/crates/core/src/messages/error.rs
@@ -15,9 +15,9 @@ pub enum Error {
     #[error(transparent)]
     Auth(#[from] litellm_auth::Error),
     #[error(transparent)]
-    Transport(#[from] litellm_llms::custom_httpx::transport::Error),
+    Transport(#[from] litellm_http::transport::Error),
     #[error(transparent)]
-    Headers(#[from] litellm_llms::custom_httpx::http_handler::HeaderError),
+    Headers(#[from] litellm_http::request::HeaderError),
 }
 
 impl From<LlmError> for Error {
diff --git a/litellm-rust/crates/core/src/messages/handler.rs b/litellm-rust/crates/core/src/messages/handler.rs
index 22e2c398ff7..fe7e8bb4b80 100644
--- a/litellm-rust/crates/core/src/messages/handler.rs
+++ b/litellm-rust/crates/core/src/messages/handler.rs
@@ -1,9 +1,7 @@
 use std::time::Duration;
 
-use litellm_llms::{
-    base_llm::anthropic_messages::transformation::BaseAnthropicMessagesConfig,
-    custom_httpx::{http_handler::http_request, transport::Error as TransportError},
-};
+use litellm_http::{request::http_request, transport::Error as TransportError};
+use litellm_llms::base_llm::anthropic_messages::transformation::BaseAnthropicMessagesConfig;
 use litellm_types::llms::anthropic_messages::anthropic_response::AnthropicMessagesResponse;
 use serde_json::Value;
 
diff --git a/litellm-rust/crates/core/src/messages/tests.rs b/litellm-rust/crates/core/src/messages/tests.rs
index 55d8ead8e8b..057b42a316c 100644
--- a/litellm-rust/crates/core/src/messages/tests.rs
+++ b/litellm-rust/crates/core/src/messages/tests.rs
@@ -82,7 +82,7 @@ fn string_headers_rejects_non_string_values() {
     let err = string_headers(Some(headers)).expect_err("non-string header rejected");
     assert_eq!(
         err,
-        Error::Headers(litellm_llms::custom_httpx::http_handler::HeaderError {
+        Error::Headers(litellm_http::request::HeaderError {
             context: "messages",
             name: "x-count".to_string(),
             actual: "number",
@@ -432,7 +432,7 @@ async fn messages_maps_provider_error_status_to_http_error() {
 
     assert!(matches!(
         err,
-        Error::Transport(litellm_llms::custom_httpx::transport::Error::Http { status: 401, .. })
+        Error::Transport(litellm_http::transport::Error::Http { status: 401, .. })
     ));
 }
 
diff --git a/litellm-rust/crates/core/src/ocr/client.rs b/litellm-rust/crates/core/src/ocr/client.rs
index c7b4751bd9e..e635f93a294 100644
--- a/litellm-rust/crates/core/src/ocr/client.rs
+++ b/litellm-rust/crates/core/src/ocr/client.rs
@@ -1,6 +1,5 @@
-use litellm_llms::{
-    base_llm::ocr::{error::Error, transformation::LiteLLMOcrResponse},
-    custom_httpx::llm_http_handler::OcrClient,
+use litellm_llms::base_llm::ocr::{
+    error::Error, handler::OcrClient, transformation::LiteLLMOcrResponse,
 };
 
 use crate::ocr::{
diff --git a/litellm-rust/crates/core/src/ocr/handler.rs b/litellm-rust/crates/core/src/ocr/handler.rs
index bbf9cfa0e02..f49976de043 100644
--- a/litellm-rust/crates/core/src/ocr/handler.rs
+++ b/litellm-rust/crates/core/src/ocr/handler.rs
@@ -1,12 +1,10 @@
 use futures_util::future::BoxFuture;
 use litellm_auth::SecretValue;
 use litellm_host::event::{MachineEvent, RawResponse, RequestContext, WireRequest};
-use litellm_llms::{
-    base_llm::ocr::{
-        error::Error,
-        transformation::{LiteLLMOcrResponse, PreparedOcrRequest},
-    },
-    custom_httpx::llm_http_handler::{CallHooks, OcrClient},
+use litellm_llms::base_llm::ocr::{
+    error::Error,
+    handler::{CallHooks, OcrClient},
+    transformation::{LiteLLMOcrResponse, PreparedOcrRequest},
 };
 use serde_json::Value;
 
diff --git a/litellm-rust/crates/core/src/ocr/provider_config.rs b/litellm-rust/crates/core/src/ocr/provider_config.rs
index 14b34ea4564..ee9ba76928d 100644
--- a/litellm-rust/crates/core/src/ocr/provider_config.rs
+++ b/litellm-rust/crates/core/src/ocr/provider_config.rs
@@ -7,13 +7,13 @@ use litellm_llms::{
     },
     base_llm::ocr::{
         error::Error,
+        handler::{self, CallHooks, OcrClient},
         transformation::{
             BaseOcrConfig, LiteLLMOcrResponse, OcrCredentialInputs, OcrDocument,
             PreparedOcrRequest, ResolvedOcrCredentials,
         },
     },
     cohere::ocr::transformation::CohereParseConfig,
-    custom_httpx::llm_http_handler::{self, CallHooks, OcrClient},
     mistral::ocr::transformation::MistralOcrConfig,
     reducto::ocr::transformation::{ReductoParseLegacyConfig, ReductoParseV3Config},
     vertex_ai::ocr::{
@@ -116,7 +116,7 @@ impl OcrConfigKind {
         request: &PreparedOcrRequest,
         hooks: &dyn CallHooks<Error>,
     ) -> Result<LiteLLMOcrResponse, Error> {
-        with_config!(self, config => llm_http_handler::ocr(&config, client, request, hooks).await)
+        with_config!(self, config => handler::ocr(&config, client, request, hooks).await)
     }
 }
 
diff --git a/litellm-rust/crates/core/src/ocr/route.rs b/litellm-rust/crates/core/src/ocr/route.rs
index bfc8c5ca965..26c9ac27102 100644
--- a/litellm-rust/crates/core/src/ocr/route.rs
+++ b/litellm-rust/crates/core/src/ocr/route.rs
@@ -6,9 +6,8 @@ use litellm_host::{
     machine::{HostChannel, HostTokenProvider, MachineFault, RouteMachine, TokenRoute},
     route::Route,
 };
-use litellm_llms::{
-    base_llm::ocr::{error::Error, transformation::LiteLLMOcrResponse},
-    custom_httpx::llm_http_handler::OcrClient,
+use litellm_llms::base_llm::ocr::{
+    error::Error, handler::OcrClient, transformation::LiteLLMOcrResponse,
 };
 
 use super::handler::perform_ocr_request;
diff --git a/litellm-rust/crates/core/src/responses/error.rs b/litellm-rust/crates/core/src/responses/error.rs
index 677db2e08de..1c940d8ed9b 100644
--- a/litellm-rust/crates/core/src/responses/error.rs
+++ b/litellm-rust/crates/core/src/responses/error.rs
@@ -11,7 +11,7 @@ pub enum Error {
     #[error(transparent)]
     Auth(#[from] litellm_auth::Error),
     #[error(transparent)]
-    Transport(#[from] litellm_llms::custom_httpx::transport::Error),
+    Transport(#[from] litellm_http::transport::Error),
     #[error(transparent)]
-    Headers(#[from] litellm_llms::custom_httpx::http_handler::HeaderError),
+    Headers(#[from] litellm_http::request::HeaderError),
 }
diff --git a/litellm-rust/crates/core/src/responses/websocket.rs b/litellm-rust/crates/core/src/responses/websocket.rs
index ccf4aa75149..f57ba65a6fb 100644
--- a/litellm-rust/crates/core/src/responses/websocket.rs
+++ b/litellm-rust/crates/core/src/responses/websocket.rs
@@ -95,9 +95,7 @@ impl ResponsesWebSocketConnection {
         timeout: Option<Duration>,
     ) -> Result<Self, Error> {
         let mut request = url.into_client_request().map_err(|error| {
-            Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-                error.to_string(),
-            ))
+            Error::Transport(litellm_http::transport::Error::Network(error.to_string()))
         })?;
         for (name, value) in headers {
             let header_name = name
@@ -110,7 +108,7 @@ impl ResponsesWebSocketConnection {
         let connect = connect_upstream(request);
         let result = match timeout {
             Some(timeout) => tokio::time::timeout(timeout, connect).await.map_err(|_| {
-                Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
+                Error::Transport(litellm_http::transport::Error::Network(
                     "Responses WebSocket connection timed out".into(),
                 ))
             })?,
@@ -118,14 +116,12 @@ impl ResponsesWebSocketConnection {
         };
         let (socket, _) = result.map_err(|error| match *error {
             tokio_tungstenite::tungstenite::Error::Http(response) => {
-                Error::Transport(litellm_llms::custom_httpx::transport::Error::Http {
+                Error::Transport(litellm_http::transport::Error::Http {
                     status: response.status().as_u16(),
                     body: String::new(),
                 })
             }
-            other => Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-                other.to_string(),
-            )),
+            other => Error::Transport(litellm_http::transport::Error::Network(other.to_string())),
         })?;
         Ok(Self {
             socket: Arc::new(Mutex::new(Some(socket))),
@@ -135,16 +131,12 @@ impl ResponsesWebSocketConnection {
     pub async fn send_text(&self, text: String) -> Result<(), Error> {
         let mut socket = self.socket.lock().await;
         let Some(socket) = socket.as_mut() else {
-            return Err(Error::Transport(
-                litellm_llms::custom_httpx::transport::Error::Network(
-                    "Responses WebSocket is closed".into(),
-                ),
-            ));
+            return Err(Error::Transport(litellm_http::transport::Error::Network(
+                "Responses WebSocket is closed".into(),
+            )));
         };
         socket.send(Message::Text(text)).await.map_err(|error| {
-            Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-                error.to_string(),
-            ))
+            Error::Transport(litellm_http::transport::Error::Network(error.to_string()))
         })
     }
 
@@ -160,9 +152,9 @@ impl ResponsesWebSocketConnection {
                 .map_err(|error| Error::InvalidResponse(error.to_string())),
             Some(Ok(Message::Close(_))) | None => Ok(None),
             Some(Ok(_)) => Ok(None),
-            Some(Err(error)) => Err(Error::Transport(
-                litellm_llms::custom_httpx::transport::Error::Network(error.to_string()),
-            )),
+            Some(Err(error)) => Err(Error::Transport(litellm_http::transport::Error::Network(
+                error.to_string(),
+            ))),
         }
     }
 
@@ -170,9 +162,7 @@ impl ResponsesWebSocketConnection {
         let mut socket = self.socket.lock().await;
         if let Some(socket) = socket.as_mut() {
             socket.close(None).await.map_err(|error| {
-                Error::Transport(litellm_llms::custom_httpx::transport::Error::Network(
-                    error.to_string(),
-                ))
+                Error::Transport(litellm_http::transport::Error::Network(error.to_string()))
             })?;
         }
         *socket = None;
diff --git a/litellm-rust/crates/core/tests/ocr.rs b/litellm-rust/crates/core/tests/ocr.rs
index e7a8fc0abc1..b999c43de8b 100644
--- a/litellm-rust/crates/core/tests/ocr.rs
+++ b/litellm-rust/crates/core/tests/ocr.rs
@@ -6,16 +6,14 @@ use litellm_host::{
     host::{Host, HostOp, HostResult},
     machine::{HostFailure, Machine, MachineStep},
 };
-use litellm_http::{HttpClientPool, HttpSettings, Resolution};
-use litellm_llms::{
-    base_llm::ocr::{
-        error::Error as OcrError,
-        transformation::{LiteLLMOcrResponse, OCR_RESPONSE_MAX_BYTES, OcrTransportConfig},
-    },
-    custom_httpx::{
-        llm_http_handler::OcrClient,
-        media::{PublicDnsResolver, UrlPolicy},
-    },
+use litellm_http::{
+    HttpClientPool, HttpSettings, Resolution,
+    media::{PublicDnsResolver, UrlPolicy},
+};
+use litellm_llms::base_llm::ocr::{
+    error::Error as OcrError,
+    handler::OcrClient,
+    transformation::{LiteLLMOcrResponse, OCR_RESPONSE_MAX_BYTES, OcrTransportConfig},
 };
 use rstest::rstest;
 use serde_json::{Value, json};
@@ -624,7 +622,7 @@ async fn read_bounded_response(response: Vec<u8>, limit: usize) -> Result<bytes:
         .unwrap();
     let result = tokio::time::timeout(
         std::time::Duration::from_secs(2),
-        litellm_llms::custom_httpx::llm_http_handler::read_response_bytes(response, limit),
+        litellm_llms::base_llm::ocr::handler::read_response_bytes(response, limit),
     )
     .await;
     server.abort();
@@ -676,10 +674,7 @@ async fn oversized_error_retains_http_status_and_bounded_diagnostics_without_dra
         .await
         .unwrap_err();
     match error {
-        OcrError::Transport(litellm_llms::custom_httpx::transport::Error::Http {
-            status,
-            body,
-        }) => {
+        OcrError::Transport(litellm_http::transport::Error::Http { status, body }) => {
             assert_eq!(status, 429);
             assert_eq!(body, prefix);
         }
diff --git a/litellm-rust/crates/core/tests/ocr/support.rs b/litellm-rust/crates/core/tests/ocr/support.rs
index b368a754656..974fa3d6655 100644
--- a/litellm-rust/crates/core/tests/ocr/support.rs
+++ b/litellm-rust/crates/core/tests/ocr/support.rs
@@ -2,9 +2,10 @@ use std::sync::{Arc, Mutex};
 
 use futures_util::future::BoxFuture;
 use litellm_host::event::WireRequest;
-use litellm_llms::{
-    base_llm::ocr::{error::Error, transformation::LiteLLMOcrResponse},
-    custom_httpx::llm_http_handler::{CallHooks, OcrClient},
+use litellm_llms::base_llm::ocr::{
+    error::Error,
+    handler::{CallHooks, OcrClient},
+    transformation::LiteLLMOcrResponse,
 };
 use serde_json::{Value, json};
 use tokio::{
diff --git a/litellm-rust/crates/http/Cargo.toml b/litellm-rust/crates/http/Cargo.toml
index b0dc7693840..4f94f37a8d5 100644
--- a/litellm-rust/crates/http/Cargo.toml
+++ b/litellm-rust/crates/http/Cargo.toml
@@ -5,13 +5,18 @@ edition.workspace = true
 license.workspace = true
 repository.workspace = true
 
+[features]
+test-support = []
+
 [dependencies]
 http.workspace = true
 litellm-core-utils.workspace = true
 hyper-util.workspace = true
 reqwest.workspace = true
 rustls.workspace = true
+serde_json.workspace = true
 thiserror.workspace = true
+tokio.workspace = true
 webpki-roots.workspace = true
 
 [dev-dependencies]
diff --git a/litellm-rust/crates/http/src/lib.rs b/litellm-rust/crates/http/src/lib.rs
index ddbc3b63b08..c6d9959348d 100644
--- a/litellm-rust/crates/http/src/lib.rs
+++ b/litellm-rust/crates/http/src/lib.rs
@@ -1,9 +1,12 @@
 mod config;
 mod error;
+pub mod media;
 mod pool;
 mod proxy;
+pub mod request;
 mod settings;
 mod tls;
+pub mod transport;
 
 pub use config::{HttpClientConfig, Resolution, Verify};
 pub use error::Error;
diff --git a/litellm-rust/crates/llms/src/custom_httpx/media.rs b/litellm-rust/crates/http/src/media.rs
similarity index 97%
rename from litellm-rust/crates/llms/src/custom_httpx/media.rs
rename to litellm-rust/crates/http/src/media.rs
index 059d0a05010..ae3f55b476a 100644
--- a/litellm-rust/crates/llms/src/custom_httpx/media.rs
+++ b/litellm-rust/crates/http/src/media.rs
@@ -7,12 +7,13 @@ use std::{
     time::Duration,
 };
 
-use litellm_http::{ClientVariant, HttpClientConfig, HttpClientPool};
 use reqwest::{
     Url,
     dns::{Addrs, Name, Resolve, Resolving},
 };
 
+use crate::{ClientVariant, HttpClientConfig, HttpClientPool};
+
 #[derive(Debug, thiserror::Error)]
 pub enum Error {
     #[error("media URL rejected by network policy")]
@@ -32,7 +33,7 @@ pub enum Error {
     #[error("media download timed out")]
     Timeout,
     #[error("{0}")]
-    Transport(#[from] crate::custom_httpx::transport::Error),
+    Transport(#[from] crate::transport::Error),
 }
 
 #[derive(Clone, Debug, PartialEq, Eq)]
@@ -101,7 +102,7 @@ impl MediaFetcher {
         pool: &HttpClientPool,
         config: &HttpClientConfig,
         url_policy: UrlPolicy,
-    ) -> Result<Self, litellm_http::Error> {
+    ) -> Result<Self, crate::Error> {
         let proxies = config.proxies.clone();
         let uses_proxy: ProxyMatch = Arc::new(move |url| proxies.apply_to(url));
         Self::with_resolution(
@@ -119,7 +120,7 @@ impl MediaFetcher {
         url_policy: UrlPolicy,
         address_resolver: Arc<dyn AddressResolver>,
         uses_proxy: ProxyMatch,
-    ) -> Result<Self, litellm_http::Error> {
+    ) -> Result<Self, crate::Error> {
         Ok(Self {
             pinned: pool.client(config, ClientVariant::Media)?,
             unpinned: pool.client(config, ClientVariant::UnpinnedMedia)?,
@@ -164,7 +165,7 @@ impl MediaFetcher {
                 .get(url.clone())
                 .send()
                 .await
-                .map_err(crate::custom_httpx::transport::Error::from)?;
+                .map_err(crate::transport::Error::from)?;
             if response.status().is_redirection() {
                 if redirects_followed == policy.max_redirects {
                     return Err(Error::TooManyRedirects);
@@ -195,7 +196,7 @@ impl MediaFetcher {
             while let Some(chunk) = response
                 .chunk()
                 .await
-                .map_err(crate::custom_httpx::transport::Error::from)?
+                .map_err(crate::transport::Error::from)?
             {
                 enforce_download_size(bytes.len() as u64 + chunk.len() as u64, policy.max_bytes)?;
                 bytes.extend_from_slice(&chunk);
@@ -245,7 +246,7 @@ impl MediaFetcher {
             .address_resolver
             .resolve(host, port)
             .await
-            .map_err(|error| crate::custom_httpx::transport::Error::Network(error.to_string()))?;
+            .map_err(|error| crate::transport::Error::Network(error.to_string()))?;
         validate_addresses(&addresses)
     }
 }
@@ -346,13 +347,13 @@ impl Resolve for PublicDnsResolver {
 mod tests {
     use std::collections::HashSet;
 
-    use litellm_http::{HttpSettings, Resolution};
     use tokio::{
         io::{AsyncReadExt, AsyncWriteExt},
         net::TcpListener,
     };
 
     use super::*;
+    use crate::{HttpSettings, Resolution};
 
     async fn serve(response: &'static [u8]) -> (Url, tokio::task::JoinHandle<()>) {
         let listener = TcpListener::bind("127.0.0.1:0")
diff --git a/litellm-rust/crates/llms/src/custom_httpx/http_handler.rs b/litellm-rust/crates/http/src/request.rs
similarity index 93%
rename from litellm-rust/crates/llms/src/custom_httpx/http_handler.rs
rename to litellm-rust/crates/http/src/request.rs
index e629be37336..874a0f3abf9 100644
--- a/litellm-rust/crates/llms/src/custom_httpx/http_handler.rs
+++ b/litellm-rust/crates/http/src/request.rs
@@ -13,20 +13,12 @@ use serde_json::{Map, Value};
 /// before truncation, so provider bodies are bounded and data-minimized.
 const UPSTREAM_ERROR_BODY_MAX_CHARS: usize = 256;
 
-#[allow(
-    dead_code,
-    reason = "used by the OCR architecture in the next stacked PR"
-)]
 pub enum HeaderPolicy<'a> {
     All,
     Only(&'a [&'a str]),
     Except(&'a [&'a str]),
 }
 
-#[allow(
-    dead_code,
-    reason = "used by the OCR architecture in the next stacked PR"
-)]
 pub fn with_headers(
     builder: reqwest::RequestBuilder,
     headers: &[(String, String)],
@@ -107,18 +99,6 @@ pub fn has_bearer_auth(headers: &[(String, String)]) -> bool {
     })
 }
 
-#[allow(
-    dead_code,
-    reason = "used by the OCR architecture in the next stacked PR"
-)]
-pub fn deserialize_optional_param<'de, D, T>(deserializer: D) -> Result<Option<Option<T>>, D::Error>
-where
-    D: serde::Deserializer<'de>,
-    T: serde::Deserialize<'de>,
-{
-    <Option<T> as serde::Deserialize>::deserialize(deserializer).map(Some)
-}
-
 #[cfg(test)]
 mod tests {
     use serde_json::json;
diff --git a/litellm-rust/crates/llms/src/custom_httpx/transport.rs b/litellm-rust/crates/http/src/transport.rs
similarity index 88%
rename from litellm-rust/crates/llms/src/custom_httpx/transport.rs
rename to litellm-rust/crates/http/src/transport.rs
index c42cdf410f6..8814925bbf2 100644
--- a/litellm-rust/crates/llms/src/custom_httpx/transport.rs
+++ b/litellm-rust/crates/http/src/transport.rs
@@ -46,11 +46,8 @@ mod tests {
             .send()
             .await
             .expect_err("invalid port");
-        let error = crate::custom_httpx::transport::Error::from_reqwest_before_dispatch(error);
-        assert!(matches!(
-            error,
-            crate::custom_httpx::transport::Error::Connect(_)
-        ));
+        let error = crate::transport::Error::from_reqwest_before_dispatch(error);
+        assert!(matches!(error, crate::transport::Error::Connect(_)));
         assert!(!error.to_string().contains("secret"));
         assert!(!error.to_string().contains("private"));
     }
@@ -76,7 +73,7 @@ mod tests {
             .await
             .expect_err("nothing listens on the port");
         let root_cause = root_cause(&error).expect("reqwest reports a cause");
-        let message = crate::custom_httpx::transport::Error::from(error).to_string();
+        let message = crate::transport::Error::from(error).to_string();
         assert!(message.contains(&root_cause), "{message}");
         assert!(!message.contains("secret"));
     }
@@ -105,8 +102,8 @@ mod tests {
         let error = response.expect_err("server does not respond");
         assert!(error.is_timeout());
         assert!(matches!(
-            crate::custom_httpx::transport::Error::from_reqwest_before_dispatch(error),
-            crate::custom_httpx::transport::Error::Network(_)
+            crate::transport::Error::from_reqwest_before_dispatch(error),
+            crate::transport::Error::Network(_)
         ));
     }
 }
diff --git a/litellm-rust/crates/llms/AGENTS.md b/litellm-rust/crates/llms/AGENTS.md
index 09fe20cd9d6..bd1c58142fd 100644
--- a/litellm-rust/crates/llms/AGENTS.md
+++ b/litellm-rust/crates/llms/AGENTS.md
@@ -1,4 +1,4 @@
-litellm-llms mirrors `litellm/llms/`: base config traits, provider transformations, and the `custom_httpx` handlers. See `../core/AGENTS.md` for how the crates layer.
+litellm-llms mirrors `litellm/llms/`: base config traits, provider transformations, and the OCR request handler in `base_llm/ocr/handler.rs`. Transport code (clients, media fetching, header helpers, transport errors) lives in `litellm-http`. See `../core/AGENTS.md` for how the crates layer.
 
 ## Python/Rust transformation pairs
 
diff --git a/litellm-rust/crates/llms/Cargo.toml b/litellm-rust/crates/llms/Cargo.toml
index a81a4427b4d..7afc4171ca8 100644
--- a/litellm-rust/crates/llms/Cargo.toml
+++ b/litellm-rust/crates/llms/Cargo.toml
@@ -6,7 +6,7 @@ license.workspace = true
 repository.workspace = true
 
 [features]
-test-support = []
+test-support = ["litellm-http/test-support"]
 
 [dependencies]
 litellm-types.workspace = true
diff --git a/litellm-rust/crates/llms/src/azure_ai/ocr/cohere_parse_transformation.rs b/litellm-rust/crates/llms/src/azure_ai/ocr/cohere_parse_transformation.rs
index f55f6b067e4..86ee0d96895 100644
--- a/litellm-rust/crates/llms/src/azure_ai/ocr/cohere_parse_transformation.rs
+++ b/litellm-rust/crates/llms/src/azure_ai/ocr/cohere_parse_transformation.rs
@@ -5,6 +5,7 @@ use crate::{
     base_llm::ocr::{
         document::{inline_remote_document, validate_inline_document},
         error::Error,
+        handler::OcrClient,
         transformation::{
             BaseOcrConfig, LiteLLMOcrResponse, OcrDocument, OcrRequestContext, OcrResponseFormat,
             PreparedOcrRequest,
@@ -13,7 +14,6 @@ use crate::{
     cohere::ocr::transformation::{
         CohereOptions, CohereParseConfig, CohereRequest, validate_document,
     },
-    custom_httpx::llm_http_handler::OcrClient,
 };
 
 #[derive(Default)]
@@ -108,7 +108,7 @@ impl BaseOcrConfig for AzureAICohereParseConfig {
     }
 
     fn validate_request_body(&self, body: &Value) -> Result<(), Error> {
-        let document = crate::custom_httpx::llm_http_handler::body_document(body)?;
+        let document = crate::base_llm::ocr::handler::body_document(body)?;
         validate_document(&document)?;
         validate_inline_document(&document)
     }
diff --git a/litellm-rust/crates/llms/src/azure_ai/ocr/document_intelligence/transformation.rs b/litellm-rust/crates/llms/src/azure_ai/ocr/document_intelligence/transformation.rs
index 2e398d0287e..a347375510d 100644
--- a/litellm-rust/crates/llms/src/azure_ai/ocr/document_intelligence/transformation.rs
+++ b/litellm-rust/crates/llms/src/azure_ai/ocr/document_intelligence/transformation.rs
@@ -14,19 +14,16 @@ use serde_json::{Map, Value};
 use serde_with::serde_as;
 use tokio::time::Instant;
 
-use crate::{
-    base_llm::ocr::{
-        document::InlineDocument,
-        error::Error,
-        transformation::{
-            BaseOcrConfig, DecodedOcrResponse, LiteLLMOcrResponse, OCR_INLINE_MAX_BYTES,
-            OCR_POLL_RETRY_SECS, OcrConnection, OcrCredentialInputs, OcrDocument, OcrPage,
-            OcrPageDimensions, OcrResponseContext, OcrResponseFormat, OcrUsageInfo,
-            PreparedOcrRequest, ResolvedOcrCredentials, credential_env,
-            decode_and_normalize_response, decode_response,
-        },
+use crate::base_llm::ocr::{
+    document::InlineDocument,
+    error::Error,
+    handler::{CallHooks, OcrClient, read_json_response},
+    transformation::{
+        BaseOcrConfig, DecodedOcrResponse, LiteLLMOcrResponse, OCR_INLINE_MAX_BYTES,
+        OCR_POLL_RETRY_SECS, OcrConnection, OcrCredentialInputs, OcrDocument, OcrPage,
+        OcrPageDimensions, OcrResponseContext, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest,
+        ResolvedOcrCredentials, credential_env, decode_and_normalize_response, decode_response,
     },
-    custom_httpx::llm_http_handler::{CallHooks, OcrClient, read_json_response},
 };
 
 const AZURE_DI_API_VERSION: &str = "2024-11-30";
@@ -440,7 +437,7 @@ async fn read_operation_response(
     hooks: &dyn CallHooks<Error>,
 ) -> Result<DecodedOcrResponse<AzureDocumentIntelligenceOperation>, Error> {
     if response.status() != reqwest::StatusCode::ACCEPTED {
-        let bytes = crate::custom_httpx::llm_http_handler::read_response_bytes(
+        let bytes = crate::base_llm::ocr::handler::read_response_bytes(
             response,
             connection.max_response_bytes,
         )
@@ -462,11 +459,9 @@ async fn read_operation_response(
     {
         return Err(Error::PollOrigin);
     }
-    let bytes = crate::custom_httpx::llm_http_handler::read_response_bytes(
-        response,
-        connection.max_response_bytes,
-    )
-    .await?;
+    let bytes =
+        crate::base_llm::ocr::handler::read_response_bytes(response, connection.max_response_bytes)
+            .await?;
     hooks.response_received(&bytes).await?;
     poll_operation(http_client, operation, headers, connection, native, hooks).await
 }
@@ -491,21 +486,19 @@ async fn poll_operation(
         let builder = http_client
             .get(url.clone())
             .timeout(remaining.min(connection.timeout));
-        let builder = crate::custom_httpx::http_handler::with_headers(
+        let builder = litellm_http::request::with_headers(
             builder,
             headers,
-            crate::custom_httpx::http_handler::HeaderPolicy::Only(&[
+            litellm_http::request::HeaderPolicy::Only(&[
                 AZURE_DI_SUBSCRIPTION_HEADER,
                 "authorization",
             ]),
         );
-        let response = tokio::time::timeout_at(
-            deadline,
-            crate::custom_httpx::http_handler::http_request(builder),
-        )
-        .await
-        .map_err(|_| Error::PollTimeout)?
-        .map_err(crate::custom_httpx::transport::Error::from)?;
+        let response =
+            tokio::time::timeout_at(deadline, litellm_http::request::http_request(builder))
+                .await
+                .map_err(|_| Error::PollTimeout)?
+                .map_err(litellm_http::transport::Error::from)?;
         let retry = response
             .headers()
             .get(reqwest::header::RETRY_AFTER)
@@ -580,8 +573,8 @@ impl AzureDocumentIntelligenceOcrConfig {
         config: &AzureAuthInputs,
         env_lookup: &(dyn Fn(&str) -> Option<String> + Sync),
     ) -> Result<Vec<(String, String)>, Error> {
-        if crate::custom_httpx::http_handler::has_header(&connection.extra_headers, "authorization")
-            || crate::custom_httpx::http_handler::has_header(
+        if litellm_http::request::has_header(&connection.extra_headers, "authorization")
+            || litellm_http::request::has_header(
                 &connection.extra_headers,
                 AZURE_DI_SUBSCRIPTION_HEADER,
             )
diff --git a/litellm-rust/crates/llms/src/azure_ai/ocr/transformation.rs b/litellm-rust/crates/llms/src/azure_ai/ocr/transformation.rs
index 7ef051e8986..4a04910aa9a 100644
--- a/litellm-rust/crates/llms/src/azure_ai/ocr/transformation.rs
+++ b/litellm-rust/crates/llms/src/azure_ai/ocr/transformation.rs
@@ -7,12 +7,12 @@ use crate::{
     base_llm::ocr::{
         document::{inline_remote_document, validate_inline_document},
         error::Error,
+        handler::OcrClient,
         transformation::{
             BaseOcrConfig, LiteLLMOcrResponse, OcrConnection, OcrDocument, OcrRequestContext,
             OcrResponseFormat, PreparedOcrRequest, credential_env,
         },
     },
-    custom_httpx::llm_http_handler::OcrClient,
     mistral::ocr::transformation::{MistralOcrConfig, MistralOcrRequest},
 };
 
@@ -107,7 +107,7 @@ impl BaseOcrConfig for AzureAiOcrConfig {
     }
 
     fn validate_request_body(&self, body: &Value) -> Result<(), Error> {
-        validate_inline_document(&crate::custom_httpx::llm_http_handler::body_document(body)?)
+        validate_inline_document(&crate::base_llm::ocr::handler::body_document(body)?)
     }
 }
 
@@ -134,8 +134,7 @@ impl AzureAiOcrConfig {
         env_lookup: &(dyn Fn(&str) -> Option<String> + Sync),
     ) -> Result<Vec<(String, String)>, Error> {
         Self::resolve_api_base(connection.api_base.as_deref(), env_lookup)?;
-        if crate::custom_httpx::http_handler::has_header(&connection.extra_headers, "authorization")
-        {
+        if litellm_http::request::has_header(&connection.extra_headers, "authorization") {
             if config.azure_ad_token_provider.is_some() {
                 super::common_utils::resolve_entra(config, env_lookup).await?;
             }
diff --git a/litellm-rust/crates/llms/src/base_llm/ocr/document.rs b/litellm-rust/crates/llms/src/base_llm/ocr/document.rs
index 8737232a075..7ff88c6b843 100644
--- a/litellm-rust/crates/llms/src/base_llm/ocr/document.rs
+++ b/litellm-rust/crates/llms/src/base_llm/ocr/document.rs
@@ -1,18 +1,14 @@
 use base64::{Engine, engine::general_purpose::STANDARD};
 use data_url::{DataUrl, DataUrlError, forgiving_base64::DecodeError, mime::Mime};
+use litellm_http::{
+    media::{DownloadPolicy, Error as MediaError, MediaFetcher},
+    transport::Error as TransportError,
+};
 use reqwest::Url;
 
-use crate::{
-    base_llm::ocr::{
-        error::Error,
-        transformation::{
-            OCR_INLINE_MAX_BYTES, OCR_MAX_FETCH_REDIRECTS, OcrConnection, OcrDocument,
-        },
-    },
-    custom_httpx::{
-        media::{DownloadPolicy, Error as MediaError, MediaFetcher},
-        transport::Error as TransportError,
-    },
+use crate::base_llm::ocr::{
+    error::Error,
+    transformation::{OCR_INLINE_MAX_BYTES, OCR_MAX_FETCH_REDIRECTS, OcrConnection, OcrDocument},
 };
 
 pub struct InlineDocument<'a>(DataUrl<'a>);
@@ -196,10 +192,8 @@ mod tests {
             .redirect(reqwest::redirect::Policy::none())
             .build()
             .unwrap();
-        let client = crate::custom_httpx::llm_http_handler::OcrClient::for_test(
-            provider_http,
-            document_http,
-        );
+        let client =
+            crate::base_llm::ocr::handler::OcrClient::for_test(provider_http, document_http);
         let converted = inline_remote_document(
             client.document_fetcher(),
             OcrDocument::ImageUrl {
diff --git a/litellm-rust/crates/llms/src/base_llm/ocr/error.rs b/litellm-rust/crates/llms/src/base_llm/ocr/error.rs
index 3061a9fe2b2..9fce387beb5 100644
--- a/litellm-rust/crates/llms/src/base_llm/ocr/error.rs
+++ b/litellm-rust/crates/llms/src/base_llm/ocr/error.rs
@@ -95,11 +95,11 @@ pub enum Error {
     #[error(transparent)]
     Auth(#[from] litellm_auth::Error),
     #[error(transparent)]
-    Transport(#[from] crate::custom_httpx::transport::Error),
+    Transport(#[from] litellm_http::transport::Error),
     #[error(transparent)]
     Params(#[from] litellm_core_utils::params::Error),
     #[error(transparent)]
-    Headers(#[from] crate::custom_httpx::http_handler::HeaderError),
+    Headers(#[from] litellm_http::request::HeaderError),
 }
 
 impl From<litellm_host::machine::MachineFault> for Error {
@@ -125,9 +125,7 @@ impl Error {
     pub fn http_status_code(&self) -> Option<u16> {
         match self {
             Self::Provider { status, .. }
-            | Self::Transport(crate::custom_httpx::transport::Error::Http { status, .. }) => {
-                Some(*status)
-            }
+            | Self::Transport(litellm_http::transport::Error::Http { status, .. }) => Some(*status),
             error if error.is_request() => Some(400),
             _ => None,
         }
diff --git a/litellm-rust/crates/llms/src/custom_httpx/llm_http_handler.rs b/litellm-rust/crates/llms/src/base_llm/ocr/handler.rs
similarity index 94%
rename from litellm-rust/crates/llms/src/custom_httpx/llm_http_handler.rs
rename to litellm-rust/crates/llms/src/base_llm/ocr/handler.rs
index 58dc03eea2d..b6f266928b1 100644
--- a/litellm-rust/crates/llms/src/custom_httpx/llm_http_handler.rs
+++ b/litellm-rust/crates/llms/src/base_llm/ocr/handler.rs
@@ -2,22 +2,20 @@ use bytes::{Bytes, BytesMut};
 use futures_util::future::BoxFuture;
 use litellm_auth_gcp::VertexAuth;
 use litellm_host::event::WireRequest;
-use litellm_http::{ClientVariant, HttpClientConfig, HttpClientPool};
+use litellm_http::{
+    ClientVariant, HttpClientConfig, HttpClientPool,
+    media::{MediaFetcher, UrlPolicy},
+    request::{HeaderPolicy, execute_http_request, with_headers},
+    transport,
+};
 use serde::{Serialize, de::DeserializeOwned};
 use serde_json::Value;
 
-use crate::{
-    base_llm::ocr::{
-        error::Error,
-        transformation::{
-            BaseOcrConfig, DecodedOcrResponse, LiteLLMOcrResponse, OcrDocument, OcrResponseContext,
-            PreparedOcrRequest, decode_request_value, decode_response,
-        },
-    },
-    custom_httpx::{
-        http_handler::{HeaderPolicy, execute_http_request, with_headers},
-        media::{MediaFetcher, UrlPolicy},
-        transport,
+use crate::base_llm::ocr::{
+    error::Error,
+    transformation::{
+        BaseOcrConfig, DecodedOcrResponse, LiteLLMOcrResponse, OcrDocument, OcrResponseContext,
+        PreparedOcrRequest, decode_request_value, decode_response,
     },
 };
 
diff --git a/litellm-rust/crates/llms/src/base_llm/ocr/mod.rs b/litellm-rust/crates/llms/src/base_llm/ocr/mod.rs
index 7194efbb203..1231633431e 100644
--- a/litellm-rust/crates/llms/src/base_llm/ocr/mod.rs
+++ b/litellm-rust/crates/llms/src/base_llm/ocr/mod.rs
@@ -1,3 +1,4 @@
 pub mod document;
 pub mod error;
+pub mod handler;
 pub mod transformation;
diff --git a/litellm-rust/crates/llms/src/base_llm/ocr/transformation.rs b/litellm-rust/crates/llms/src/base_llm/ocr/transformation.rs
index f215546849d..be4551709a1 100644
--- a/litellm-rust/crates/llms/src/base_llm/ocr/transformation.rs
+++ b/litellm-rust/crates/llms/src/base_llm/ocr/transformation.rs
@@ -12,11 +12,9 @@ use serde::{
 use serde_json::{Map, Value};
 use serde_with::serde_as;
 
-use crate::{
-    base_llm::ocr::error::Error,
-    custom_httpx::llm_http_handler::{
-        CallHooks, OcrClient, read_response_bytes, transform_request_body,
-    },
+use crate::base_llm::ocr::{
+    error::Error,
+    handler::{CallHooks, OcrClient, read_response_bytes, transform_request_body},
 };
 
 pub const OCR_RESPONSE_MAX_BYTES: usize = 64 * 1024 * 1024;
diff --git a/litellm-rust/crates/llms/src/cohere/ocr/transformation.rs b/litellm-rust/crates/llms/src/cohere/ocr/transformation.rs
index 2528c967f41..da6cf90ffcf 100644
--- a/litellm-rust/crates/llms/src/cohere/ocr/transformation.rs
+++ b/litellm-rust/crates/llms/src/cohere/ocr/transformation.rs
@@ -7,17 +7,15 @@ use serde::{Deserialize, Serialize};
 use serde_json::{Map, Value};
 use serde_with::serde_as;
 
-use crate::{
-    base_llm::ocr::{
-        document::InlineDocument,
-        error::Error,
-        transformation::{
-            BaseOcrConfig, LiteLLMOcrResponse, OCR_INLINE_MAX_BYTES, OcrConnection, OcrDocument,
-            OcrPage, OcrPageImage, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest,
-            credential_env, decode_and_normalize_response, decode_response_value,
-        },
+use crate::base_llm::ocr::{
+    document::InlineDocument,
+    error::Error,
+    handler::OcrClient,
+    transformation::{
+        BaseOcrConfig, LiteLLMOcrResponse, OCR_INLINE_MAX_BYTES, OcrConnection, OcrDocument,
+        OcrPage, OcrPageImage, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest, credential_env,
+        decode_and_normalize_response, decode_response_value,
     },
-    custom_httpx::llm_http_handler::OcrClient,
 };
 
 const COHERE_PARSE_API_BASE: &str = "https://api.cohere.com";
@@ -163,7 +161,7 @@ impl BaseOcrConfig for CohereParseConfig {
     }
 
     fn validate_request_body(&self, body: &Value) -> Result<(), Error> {
-        validate_document(&crate::custom_httpx::llm_http_handler::body_document(body)?)
+        validate_document(&crate::base_llm::ocr::handler::body_document(body)?)
     }
 }
 
@@ -173,8 +171,7 @@ impl CohereParseConfig {
         connection: &OcrConnection,
         env_lookup: &(dyn Fn(&str) -> Option<String> + Sync),
     ) -> Result<Vec<(String, String)>, Error> {
-        if crate::custom_httpx::http_handler::has_header(&connection.extra_headers, "authorization")
-        {
+        if litellm_http::request::has_header(&connection.extra_headers, "authorization") {
             return Ok(connection.extra_headers.clone());
         }
         let key = connection
diff --git a/litellm-rust/crates/llms/src/custom_httpx/mod.rs b/litellm-rust/crates/llms/src/custom_httpx/mod.rs
deleted file mode 100644
index 057cb796c09..00000000000
--- a/litellm-rust/crates/llms/src/custom_httpx/mod.rs
+++ /dev/null
@@ -1,4 +0,0 @@
-pub mod http_handler;
-pub mod llm_http_handler;
-pub mod media;
-pub mod transport;
diff --git a/litellm-rust/crates/llms/src/lib.rs b/litellm-rust/crates/llms/src/lib.rs
index 884fa739992..8d1bb366ed4 100644
--- a/litellm-rust/crates/llms/src/lib.rs
+++ b/litellm-rust/crates/llms/src/lib.rs
@@ -3,7 +3,6 @@ pub mod azure_ai;
 pub mod base_llm;
 pub mod bedrock;
 pub mod cohere;
-pub mod custom_httpx;
 pub mod mistral;
 pub mod openai;
 pub mod reducto;
diff --git a/litellm-rust/crates/llms/src/mistral/ocr/transformation.rs b/litellm-rust/crates/llms/src/mistral/ocr/transformation.rs
index 9028f09c5ab..95658837fc3 100644
--- a/litellm-rust/crates/llms/src/mistral/ocr/transformation.rs
+++ b/litellm-rust/crates/llms/src/mistral/ocr/transformation.rs
@@ -2,16 +2,13 @@ use litellm_core_utils::{call_arguments::CallArguments, params::OpaqueParams, ur
 use serde::{Deserialize, Serialize};
 use serde_json::Value;
 
-use crate::{
-    base_llm::ocr::{
-        error::Error,
-        transformation::{
-            BaseOcrConfig, LiteLLMOcrResponse, OcrConnection, OcrDocument, OcrPage,
-            OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest, credential_env,
-            decode_and_normalize_response,
-        },
+use crate::base_llm::ocr::{
+    error::Error,
+    handler::OcrClient,
+    transformation::{
+        BaseOcrConfig, LiteLLMOcrResponse, OcrConnection, OcrDocument, OcrPage, OcrResponseFormat,
+        OcrUsageInfo, PreparedOcrRequest, credential_env, decode_and_normalize_response,
     },
-    custom_httpx::llm_http_handler::OcrClient,
 };
 
 const MISTRAL_OCR_API_BASE: &str = "https://api.mistral.ai/v1";
@@ -129,8 +126,7 @@ impl MistralOcrConfig {
         connection: &OcrConnection,
         env_lookup: &(dyn Fn(&str) -> Option<String> + Sync),
     ) -> Result<Vec<(String, String)>, Error> {
-        if crate::custom_httpx::http_handler::has_header(&connection.extra_headers, "authorization")
-        {
+        if litellm_http::request::has_header(&connection.extra_headers, "authorization") {
             return Ok(connection.extra_headers.clone());
         }
         let api_key = connection
diff --git a/litellm-rust/crates/llms/src/reducto/ocr/transformation.rs b/litellm-rust/crates/llms/src/reducto/ocr/transformation.rs
index ec876fafb8f..740f0ced090 100644
--- a/litellm-rust/crates/llms/src/reducto/ocr/transformation.rs
+++ b/litellm-rust/crates/llms/src/reducto/ocr/transformation.rs
@@ -8,18 +8,14 @@ use litellm_core_utils::{
 use serde::{Deserialize, Deserializer, Serialize};
 use serde_json::{Map, Value, json};
 
-use crate::{
-    base_llm::ocr::{
-        document::InlineDocument,
-        error::Error,
-        transformation::{
-            BaseOcrConfig, LiteLLMOcrResponse, OCR_INLINE_MAX_BYTES, OcrConnection, OcrDocument,
-            OcrPage, OcrRequestContext, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest,
-            credential_env, decode_and_normalize_response,
-        },
-    },
-    custom_httpx::llm_http_handler::{
-        CallHooks, OcrClient, build_http_request, guardrail_document,
+use crate::base_llm::ocr::{
+    document::InlineDocument,
+    error::Error,
+    handler::{CallHooks, OcrClient, build_http_request, guardrail_document},
+    transformation::{
+        BaseOcrConfig, LiteLLMOcrResponse, OCR_INLINE_MAX_BYTES, OcrConnection, OcrDocument,
+        OcrPage, OcrRequestContext, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest,
+        credential_env, decode_and_normalize_response,
     },
 };
 
@@ -437,7 +433,7 @@ fn resolve_headers(
     connection: &OcrConnection,
     env_lookup: &(dyn Fn(&str) -> Option<String> + Sync),
 ) -> Result<Vec<(String, String)>, Error> {
-    if crate::custom_httpx::http_handler::has_header(&connection.extra_headers, "authorization") {
+    if litellm_http::request::has_header(&connection.extra_headers, "authorization") {
         return Ok(connection.extra_headers.clone());
     }
     let api_key = connection
@@ -515,25 +511,21 @@ async fn upload_bytes_async(
         )?)
         .multipart(reqwest::multipart::Form::new().part("file", part))
         .timeout(connection.timeout);
-    let builder = crate::custom_httpx::http_handler::with_headers(
+    let builder = litellm_http::request::with_headers(
         builder,
         headers,
-        crate::custom_httpx::http_handler::HeaderPolicy::Except(&[
-            "content-type",
-            "content-length",
-        ]),
+        litellm_http::request::HeaderPolicy::Except(&["content-type", "content-length"]),
     );
-    let response = crate::custom_httpx::http_handler::http_request(builder)
+    let response = litellm_http::request::http_request(builder)
         .await
-        .map_err(crate::custom_httpx::transport::Error::from)?;
-    let uploaded =
-        crate::custom_httpx::llm_http_handler::read_json_response::<ReductoUploadResponse>(
-            response,
-            false,
-            connection.max_response_bytes,
-        )
-        .await?
-        .data;
+        .map_err(litellm_http::transport::Error::from)?;
+    let uploaded = crate::base_llm::ocr::handler::read_json_response::<ReductoUploadResponse>(
+        response,
+        false,
+        connection.max_response_bytes,
+    )
+    .await?
+    .data;
     let file_id = uploaded
         .file_id
         .as_deref()
diff --git a/litellm-rust/crates/llms/src/vertex_ai/ocr/deepseek_transformation.rs b/litellm-rust/crates/llms/src/vertex_ai/ocr/deepseek_transformation.rs
index 588b5243004..8009a65ff77 100644
--- a/litellm-rust/crates/llms/src/vertex_ai/ocr/deepseek_transformation.rs
+++ b/litellm-rust/crates/llms/src/vertex_ai/ocr/deepseek_transformation.rs
@@ -4,16 +4,14 @@ use serde::{Deserialize, Serialize};
 use serde_json::{Map, Value};
 
 use super::transformation::VertexAiOcrConfig;
-use crate::{
-    base_llm::ocr::{
-        error::Error,
-        transformation::{
-            BaseOcrConfig, LiteLLMOcrResponse, OcrDocument, OcrPage, OcrPageDimensions,
-            OcrPageImage, OcrRequestContext, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest,
-            credential_env, decode_and_normalize_response, decode_response_value,
-        },
+use crate::base_llm::ocr::{
+    error::Error,
+    handler::OcrClient,
+    transformation::{
+        BaseOcrConfig, LiteLLMOcrResponse, OcrDocument, OcrPage, OcrPageDimensions, OcrPageImage,
+        OcrRequestContext, OcrResponseFormat, OcrUsageInfo, PreparedOcrRequest, credential_env,
+        decode_and_normalize_response, decode_response_value,
     },
-    custom_httpx::llm_http_handler::OcrClient,
 };
 
 const DEFAULT_API_BASE: &str = "https://aiplatform.googleapis.com";
diff --git a/litellm-rust/crates/llms/src/vertex_ai/ocr/transformation.rs b/litellm-rust/crates/llms/src/vertex_ai/ocr/transformation.rs
index c2cb23d0010..a50e8261aa3 100644
--- a/litellm-rust/crates/llms/src/vertex_ai/ocr/transformation.rs
+++ b/litellm-rust/crates/llms/src/vertex_ai/ocr/transformation.rs
@@ -7,12 +7,12 @@ use crate::{
     base_llm::ocr::{
         document::{inline_remote_document, validate_inline_document},
         error::Error,
+        handler::OcrClient,
         transformation::{
             BaseOcrConfig, LiteLLMOcrResponse, OcrConnection, OcrDocument, OcrEnvironment,
             OcrRequestContext, OcrResponseFormat, PreparedOcrRequest, credential_env,
         },
     },
-    custom_httpx::llm_http_handler::OcrClient,
     mistral::ocr::transformation::{MistralOcrConfig, MistralOcrRequest},
 };
 
@@ -112,7 +112,7 @@ impl BaseOcrConfig for VertexAiOcrConfig {
     }
 
     fn validate_request_body(&self, body: &Value) -> Result<(), Error> {
-        validate_inline_document(&crate::custom_httpx::llm_http_handler::body_document(body)?)
+        validate_inline_document(&crate::base_llm::ocr::handler::body_document(body)?)
     }
 }
 
diff --git a/litellm-rust/crates/python-bridge/src/errors.rs b/litellm-rust/crates/python-bridge/src/errors.rs
index 61c5947ed9e..19d28f76b6f 100644
--- a/litellm-rust/crates/python-bridge/src/errors.rs
+++ b/litellm-rust/crates/python-bridge/src/errors.rs
@@ -1,7 +1,6 @@
 use litellm_core::{Error, audio_transcription, chat_completions, messages, responses};
-use litellm_llms::{
-    base_llm::ocr::error::Error as OcrError, custom_httpx::transport::Error as TransportError,
-};
+use litellm_http::transport::Error as TransportError;
+use litellm_llms::base_llm::ocr::error::Error as OcrError;
 use pyo3::{
     exceptions::{PyRuntimeError, PyValueError},
     prelude::*,
diff --git a/litellm-rust/crates/python-bridge/src/http.rs b/litellm-rust/crates/python-bridge/src/http.rs
index 1fc3e4a60f1..7e9a5f093b4 100644
--- a/litellm-rust/crates/python-bridge/src/http.rs
+++ b/litellm-rust/crates/python-bridge/src/http.rs
@@ -8,8 +8,8 @@ use litellm_core_utils::settings::ProcessEnvironment;
 use litellm_http::{
     HttpClientConfig, HttpClientPool, HttpSettings, HttpSettingsLayer, Resolution, SslVerify,
     Unsupported,
+    media::{PublicDnsResolver, UrlPolicy},
 };
-use litellm_llms::custom_httpx::media::{PublicDnsResolver, UrlPolicy};
 use pyo3::{prelude::*, types::PyDict};
 
 use crate::{errors::RustBridgeDeclined, python_settings::PythonSettings};
diff --git a/litellm-rust/crates/python-bridge/src/routes/messages/host.rs b/litellm-rust/crates/python-bridge/src/routes/messages/host.rs
index c1b3f59df58..1a9b170f661 100644
--- a/litellm-rust/crates/python-bridge/src/routes/messages/host.rs
+++ b/litellm-rust/crates/python-bridge/src/routes/messages/host.rs
@@ -4,7 +4,7 @@ use litellm_core::messages::{
     route::{Messages, MessagesCall, MessagesOp, MessagesOpResult, MessagesOutput},
 };
 use litellm_host_python::{InvokeError, RouteHost, from_py, lookup, to_py};
-use litellm_llms::custom_httpx::transport::Error as TransportError;
+use litellm_http::transport::Error as TransportError;
 use pyo3::{
     exceptions::{PyException, PyValueError},
     gc::{PyTraverseError, PyVisit},
diff --git a/litellm-rust/crates/python-bridge/src/routes/ocr/errors.rs b/litellm-rust/crates/python-bridge/src/routes/ocr/errors.rs
index 0ae56efbf02..b0a6acdebfd 100644
--- a/litellm-rust/crates/python-bridge/src/routes/ocr/errors.rs
+++ b/litellm-rust/crates/python-bridge/src/routes/ocr/errors.rs
@@ -15,10 +15,9 @@ pub(super) fn to_pyerr(error: Error) -> PyErr {
                 body,
                 headers,
             } => upstream_error(py, status, body, headers)?,
-            Error::Transport(litellm_llms::custom_httpx::transport::Error::Http {
-                status,
-                body,
-            }) => upstream_error(py, status, body, Vec::new())?,
+            Error::Transport(litellm_http::transport::Error::Http { status, body }) => {
+                upstream_error(py, status, body, Vec::new())?
+            }
             Error::RequestFormat => {
                 let error = core_error_to_pyerr(Error::RequestFormat.into());
                 error
diff --git a/litellm-rust/crates/python-bridge/src/routes/ocr/mod.rs b/litellm-rust/crates/python-bridge/src/routes/ocr/mod.rs
index f9d7024c824..190f37d075d 100644
--- a/litellm-rust/crates/python-bridge/src/routes/ocr/mod.rs
+++ b/litellm-rust/crates/python-bridge/src/routes/ocr/mod.rs
@@ -9,7 +9,7 @@ use host::OcrRouteHost;
 use litellm_auth_gcp::VertexAuth;
 use litellm_callbacks_legacy::{LegacySurface, PublicCall, run_legacy_call};
 use litellm_core::ocr::route::ocr_machine;
-use litellm_llms::custom_httpx::llm_http_handler::OcrClient;
+use litellm_llms::base_llm::ocr::handler::OcrClient;
 use pyo3::{
     prelude::*,
     types::{PyDict, PyTuple},