vlm plus useafterfree fix ughhh

Author: VyasGuru

examples/flutter/RunAnywhereAI/lib/app/runanywhere_ai_app.dart

@@ -12,6 +12,7 @@ import 'package:runanywhere_ai/core/utilities/constants.dart';
 import 'package:runanywhere_ai/core/utilities/keychain_helper.dart';
 import 'package:runanywhere/public/extensions/rag_module.dart';
 import 'package:runanywhere_llamacpp/runanywhere_llamacpp.dart';
+import 'package:runanywhere_onnx/runanywhere_onnx.dart';
 
 /// RunAnywhereAIApp (mirroring iOS RunAnywhereAIApp.swift)
 ///
@@ -286,6 +287,14 @@ class _RunAnywhereAIAppState extends State<RunAnywhereAIApp> {
     debugPrint('✅ ONNX Embedding models registered');
     await Future<void>.delayed(Duration.zero);
 
+    // --- ONNX BACKEND (required for embeddings used by RAG) ---
+    try {
+      await Onnx.register();
+      debugPrint('✅ ONNX backend registered (STT + TTS + VAD + Embeddings)');
+    } catch (e) {
+      debugPrint('⚠️ ONNX backend not available: $e');
+    }
+
     // --- RAG BACKEND ---
     try {
       await RAGModule.register();

examples/flutter/RunAnywhereAI/lib/features/vision/vlm_view_model.dart

@@ -161,7 +161,7 @@ class VLMViewModel extends ChangeNotifier {
       final result = await sdk.RunAnywhere.processImageStream(
         image,
         prompt: 'Describe what you see briefly.',
-        maxTokens: 200,
+        options: const sdk.VLMGenerationOptions(maxTokens: 200),
       );
 
       // Listen to stream and append tokens
@@ -201,7 +201,7 @@ class VLMViewModel extends ChangeNotifier {
       final result = await sdk.RunAnywhere.processImageStream(
         image,
         prompt: 'Describe this image in detail.',
-        maxTokens: 300,
+        options: const sdk.VLMGenerationOptions(maxTokens: 300),
       );
 
       // Listen to stream and append tokens
@@ -285,7 +285,7 @@ class VLMViewModel extends ChangeNotifier {
       final result = await sdk.RunAnywhere.processImageStream(
         image,
         prompt: 'Describe what you see in one sentence.',
-        maxTokens: 100,
+        options: const sdk.VLMGenerationOptions(maxTokens: 100),
       );
 
       // Listen to stream and build description

examples/flutter/RunAnywhereAI/pubspec.yaml

@@ -17,6 +17,10 @@ dependencies:
   # RunAnywhere SDK - LlamaCpp Backend (LLM)
   runanywhere_llamacpp:
     path: ../../../sdk/runanywhere-flutter/packages/runanywhere_llamacpp
+
+  # RunAnywhere SDK - ONNX Backend (STT/TTS/VAD/Embeddings)
+  runanywhere_onnx:
+    path: ../../../sdk/runanywhere-flutter/packages/runanywhere_onnx
   provider: ^6.1.0
   flutter_markdown: ^0.6.18
   record: ^6.1.0

sdk/runanywhere-commons/include/rac/infrastructure/network/rac_auth_manager.h

@@ -96,12 +96,12 @@ typedef struct {
  * @brief Initialize auth manager
  * @param storage Secure storage callbacks (can be NULL for in-memory only)
  */
-void rac_auth_init(const rac_secure_storage_t* storage);
+RAC_API void rac_auth_init(const rac_secure_storage_t* storage);
 
 /**
  * @brief Reset auth manager state
  */
-void rac_auth_reset(void);
+RAC_API void rac_auth_reset(void);
 
 // =============================================================================
 // Token State
@@ -111,7 +111,7 @@ void rac_auth_reset(void);
  * @brief Check if currently authenticated
  * @return true if valid access token exists
  */
-bool rac_auth_is_authenticated(void);
+RAC_API bool rac_auth_is_authenticated(void);
 
 /**
  * @brief Check if token needs refresh
@@ -120,31 +120,31 @@ bool rac_auth_is_authenticated(void);
  *
  * @return true if token should be refreshed
  */
-bool rac_auth_needs_refresh(void);
+RAC_API bool rac_auth_needs_refresh(void);
 
 /**
  * @brief Get current access token
  * @return Access token string, or NULL if not authenticated
  */
-const char* rac_auth_get_access_token(void);
+RAC_API const char* rac_auth_get_access_token(void);
 
 /**
  * @brief Get current device ID
  * @return Device ID string, or NULL if not set
  */
-const char* rac_auth_get_device_id(void);
+RAC_API const char* rac_auth_get_device_id(void);
 
 /**
  * @brief Get current user ID
  * @return User ID string, or NULL if not set
  */
-const char* rac_auth_get_user_id(void);
+RAC_API const char* rac_auth_get_user_id(void);
 
 /**
  * @brief Get current organization ID
  * @return Organization ID string, or NULL if not set
  */
-const char* rac_auth_get_organization_id(void);
+RAC_API const char* rac_auth_get_organization_id(void);
 
 // =============================================================================
 // Request Building
@@ -158,7 +158,7 @@ const char* rac_auth_get_organization_id(void);
  * @param config SDK configuration with credentials
  * @return JSON string (caller must free), or NULL on error
  */
-char* rac_auth_build_authenticate_request(const rac_sdk_config_t* config);
+RAC_API char* rac_auth_build_authenticate_request(const rac_sdk_config_t* config);
 
 /**
  * @brief Build token refresh request JSON
@@ -167,7 +167,7 @@ char* rac_auth_build_authenticate_request(const rac_sdk_config_t* config);
  *
  * @return JSON string (caller must free), or NULL if no refresh token
  */
-char* rac_auth_build_refresh_request(void);
+RAC_API char* rac_auth_build_refresh_request(void);
 
 // =============================================================================
 // Response Handling
@@ -181,7 +181,7 @@ char* rac_auth_build_refresh_request(void);
  * @param json JSON response body
  * @return 0 on success, -1 on parse error
  */
-int rac_auth_handle_authenticate_response(const char* json);
+RAC_API int rac_auth_handle_authenticate_response(const char* json);
 
 /**
  * @brief Parse and store refresh response
@@ -191,7 +191,7 @@ int rac_auth_handle_authenticate_response(const char* json);
  * @param json JSON response body
  * @return 0 on success, -1 on parse error
  */
-int rac_auth_handle_refresh_response(const char* json);
+RAC_API int rac_auth_handle_refresh_response(const char* json);
 
 // =============================================================================
 // Token Management
@@ -214,14 +214,14 @@ int rac_auth_handle_refresh_response(const char* json);
  * @param out_needs_refresh Set to true if refresh HTTP call is needed
  * @return 0 on success (token valid), 1 if refresh needed, -1 on error
  */
-int rac_auth_get_valid_token(const char** out_token, bool* out_needs_refresh);
+RAC_API int rac_auth_get_valid_token(const char** out_token, bool* out_needs_refresh);
 
 /**
  * @brief Clear all authentication state
  *
  * Clears in-memory state and secure storage.
  */
-void rac_auth_clear(void);
+RAC_API void rac_auth_clear(void);
 
 // =============================================================================
 // Persistence
@@ -234,7 +234,7 @@ void rac_auth_clear(void);
  *
  * @return 0 on success (tokens loaded), -1 if not found or error
  */
-int rac_auth_load_stored_tokens(void);
+RAC_API int rac_auth_load_stored_tokens(void);
 
 /**
  * @brief Save current tokens to secure storage
@@ -243,7 +243,7 @@ int rac_auth_load_stored_tokens(void);
  *
  * @return 0 on success, -1 on error
  */
-int rac_auth_save_tokens(void);
+RAC_API int rac_auth_save_tokens(void);
 
 #ifdef __cplusplus
 }

sdk/runanywhere-commons/include/rac/infrastructure/network/rac_environment.h

@@ -79,56 +79,56 @@ typedef struct {
  * @param env The environment to check
  * @return true for staging/production, false for development
  */
-bool rac_env_requires_auth(rac_environment_t env);
+RAC_API bool rac_env_requires_auth(rac_environment_t env);
 
 /**
  * @brief Check if environment requires a backend URL
  * @param env The environment to check
  * @return true for staging/production, false for development
  */
-bool rac_env_requires_backend_url(rac_environment_t env);
+RAC_API bool rac_env_requires_backend_url(rac_environment_t env);
 
 /**
  * @brief Check if environment is production
  * @param env The environment to check
  * @return true only for production
  */
-bool rac_env_is_production(rac_environment_t env);
+RAC_API bool rac_env_is_production(rac_environment_t env);
 
 /**
  * @brief Check if environment is a testing environment
  * @param env The environment to check
  * @return true for development and staging
  */
-bool rac_env_is_testing(rac_environment_t env);
+RAC_API bool rac_env_is_testing(rac_environment_t env);
 
 /**
  * @brief Get the default log level for an environment
  * @param env The environment
  * @return DEBUG for development, INFO for staging, WARNING for production
  */
-rac_log_level_t rac_env_default_log_level(rac_environment_t env);
+RAC_API rac_log_level_t rac_env_default_log_level(rac_environment_t env);
 
 /**
  * @brief Check if telemetry should be sent for this environment
  * @param env The environment
  * @return true only for production
  */
-bool rac_env_should_send_telemetry(rac_environment_t env);
+RAC_API bool rac_env_should_send_telemetry(rac_environment_t env);
 
 /**
  * @brief Check if environment should sync with backend
  * @param env The environment
  * @return true for staging/production, false for development
  */
-bool rac_env_should_sync_with_backend(rac_environment_t env);
+RAC_API bool rac_env_should_sync_with_backend(rac_environment_t env);
 
 /**
  * @brief Get human-readable environment description
  * @param env The environment
  * @return String like "Development Environment"
  */
-const char* rac_env_description(rac_environment_t env);
+RAC_API const char* rac_env_description(rac_environment_t env);
 
 // =============================================================================
 // Validation Functions
@@ -155,29 +155,29 @@ typedef enum {
  * @param env The target environment
  * @return RAC_VALIDATION_OK if valid, error code otherwise
  */
-rac_validation_result_t rac_validate_api_key(const char* api_key, rac_environment_t env);
+RAC_API rac_validation_result_t rac_validate_api_key(const char* api_key, rac_environment_t env);
 
 /**
  * @brief Validate base URL for the given environment
  * @param url The URL to validate (can be NULL)
  * @param env The target environment
  * @return RAC_VALIDATION_OK if valid, error code otherwise
  */
-rac_validation_result_t rac_validate_base_url(const char* url, rac_environment_t env);
+RAC_API rac_validation_result_t rac_validate_base_url(const char* url, rac_environment_t env);
 
 /**
  * @brief Validate complete SDK configuration
  * @param config The configuration to validate
  * @return RAC_VALIDATION_OK if valid, first error code otherwise
  */
-rac_validation_result_t rac_validate_config(const rac_sdk_config_t* config);
+RAC_API rac_validation_result_t rac_validate_config(const rac_sdk_config_t* config);
 
 /**
  * @brief Get error message for validation result
  * @param result The validation result code
  * @return Human-readable error message
  */
-const char* rac_validation_error_message(rac_validation_result_t result);
+RAC_API const char* rac_validation_error_message(rac_validation_result_t result);
 
 // =============================================================================
 // Global SDK State

sdk/runanywhere-commons/scripts/build-ios.sh

@@ -520,11 +520,12 @@ create_backend_xcframework() {
             local LLAMA_BUILD="${PLATFORM_DIR}/src/backends/llamacpp/_deps/llamacpp-build"
             [[ ! -d "$LLAMA_BUILD" ]] && LLAMA_BUILD="${PLATFORM_DIR}/_deps/llamacpp-build"
 
-            for lib in llama common ggml ggml-base ggml-cpu ggml-metal ggml-blas; do
+            for lib in llama common cpp-httplib ggml ggml-base ggml-cpu ggml-metal ggml-blas; do
                 local lib_path=""
                 for possible in \
                     "${LLAMA_BUILD}/src/lib${lib}.a" \
                     "${LLAMA_BUILD}/common/lib${lib}.a" \
+                    "${LLAMA_BUILD}/vendor/cpp-httplib/lib${lib}.a" \
                     "${LLAMA_BUILD}/ggml/src/lib${lib}.a" \
                     "${LLAMA_BUILD}/ggml/src/ggml-metal/lib${lib}.a" \
                     "${LLAMA_BUILD}/ggml/src/ggml-blas/lib${lib}.a" \