diff --git a/src/openhuman/inference/provider/compatible.rs b/src/openhuman/inference/provider/compatible.rs
index d17bf8dd17..dccfea5fe1 100644
--- a/src/openhuman/inference/provider/compatible.rs
+++ b/src/openhuman/inference/provider/compatible.rs
@@ -490,6 +490,13 @@ impl OpenAiCompatibleProvider {
                     Some(model),
                     status,
                 );
+            } else if super::is_provider_config_rejection_http(status, self.name.as_str(), &error) {
+                super::log_provider_config_rejection(
+                    "responses_api",
+                    self.name.as_str(),
+                    Some(model),
+                    status,
+                );
             } else if super::should_report_provider_http_failure(status) {
                 crate::core::observability::report_error(
                     message.as_str(),
@@ -856,6 +863,13 @@ impl OpenAiCompatibleProvider {
                     Some(native_request.model.as_str()),
                     status,
                 );
+            } else if super::is_provider_config_rejection_http(status, self.name.as_str(), &body) {
+                super::log_provider_config_rejection(
+                    "streaming_chat",
+                    self.name.as_str(),
+                    Some(native_request.model.as_str()),
+                    status,
+                );
             } else if super::should_report_provider_http_failure(status) {
                 crate::core::observability::report_error(
                     message.as_str(),
@@ -1348,6 +1362,13 @@ impl Provider for OpenAiCompatibleProvider {
                     Some(model),
                     status,
                 );
+            } else if super::is_provider_config_rejection_http(status, self.name.as_str(), &error) {
+                super::log_provider_config_rejection(
+                    "chat_completions",
+                    self.name.as_str(),
+                    Some(model),
+                    status,
+                );
             } else if super::should_report_provider_http_failure(status) {
                 crate::core::observability::report_error(
                     message.as_str(),
@@ -1797,6 +1818,13 @@ impl Provider for OpenAiCompatibleProvider {
                     Some(model),
                     status,
                 );
+            } else if super::is_provider_config_rejection_http(status, self.name.as_str(), &error) {
+                super::log_provider_config_rejection(
+                    "native_chat",
+                    self.name.as_str(),
+                    Some(model),
+                    status,
+                );
             } else if super::should_report_provider_http_failure(status) {
                 crate::core::observability::report_error(
                     message.as_str(),
@@ -1952,6 +1980,17 @@ impl Provider for OpenAiCompatibleProvider {
                         Some(model_owned.as_str()),
                         status,
                     );
+                } else if super::is_provider_config_rejection_http(
+                    status,
+                    provider_name.as_str(),
+                    &raw_error,
+                ) {
+                    super::log_provider_config_rejection(
+                        "stream_chat",
+                        provider_name.as_str(),
+                        Some(model_owned.as_str()),
+                        status,
+                    );
                 } else if super::should_report_provider_http_failure(status) {
                     crate::core::observability::report_error(
                         message.as_str(),
diff --git a/src/openhuman/inference/provider/compatible_tests.rs b/src/openhuman/inference/provider/compatible_tests.rs
index 7285dac030..fd7d0266e7 100644
--- a/src/openhuman/inference/provider/compatible_tests.rs
+++ b/src/openhuman/inference/provider/compatible_tests.rs
@@ -1,4 +1,8 @@
 use super::*;
+use sentry::test::TestTransport;
+use std::sync::Arc;
+use wiremock::matchers::{method, path};
+use wiremock::{Mock, MockServer, ResponseTemplate};
 
 fn make_provider(name: &str, url: &str, key: Option<&str>) -> OpenAiCompatibleProvider {
     OpenAiCompatibleProvider::new(name, url, key, AuthStyle::Bearer)
@@ -374,6 +378,69 @@ async fn chat_via_responses_requires_non_system_message() {
         .contains("requires at least one non-system message"));
 }
 
+#[tokio::test]
+async fn streaming_chat_config_rejection_propagates_error_without_sentry_report() {
+    // Representative guardrail for the new provider-config-rejection
+    // suppression branches in compatible.rs: streaming_chat should still
+    // return an error, but it must not call report_error/Sentry for this
+    // deterministic user-config state.
+    let mock_server = MockServer::start().await;
+    Mock::given(method("POST"))
+        .and(path("/chat/completions"))
+        .respond_with(
+            ResponseTemplate::new(400)
+                .set_body_string("invalid temperature: only 1 is allowed for this model"),
+        )
+        .mount(&mock_server)
+        .await;
+
+    let transport = TestTransport::new();
+    let sentry_options = sentry::ClientOptions {
+        dsn: Some("https://public@sentry.invalid/1".parse().unwrap()),
+        transport: Some(Arc::new(transport.clone())),
+        ..Default::default()
+    };
+    let sentry_hub = Arc::new(sentry::Hub::new(
+        Some(Arc::new(sentry_options.into())),
+        Arc::new(Default::default()),
+    ));
+    let _sentry_guard = sentry::HubSwitchGuard::new(sentry_hub);
+
+    let provider =
+        OpenAiCompatibleProvider::new("custom_openai", &mock_server.uri(), None, AuthStyle::None);
+    let request = NativeChatRequest {
+        model: "kimi-k2".to_string(),
+        messages: vec![NativeMessage {
+            role: "user".to_string(),
+            content: Some("hello".to_string()),
+            tool_call_id: None,
+            tool_calls: None,
+        }],
+        temperature: Some(0.7),
+        stream: Some(true),
+        tools: None,
+        tool_choice: None,
+        thread_id: None,
+        stream_options: Some(super::compatible_types::OpenAiStreamOptions {
+            include_usage: true,
+        }),
+    };
+    let (delta_tx, _delta_rx) = tokio::sync::mpsc::channel(8);
+
+    let err = provider
+        .stream_native_chat(None, &request, &delta_tx, 0)
+        .await
+        .expect_err("400 provider config-rejection must still propagate as Err");
+    assert!(
+        err.to_string().contains("streaming API error"),
+        "err: {err}"
+    );
+    assert!(
+        transport.fetch_and_clear_events().is_empty(),
+        "provider config-rejection must not be reported to Sentry"
+    );
+}
+
 // ----------------------------------------------------------
 // Custom endpoint path tests (Issue #114)
 // ----------------------------------------------------------