diff --git a/README.md b/README.md
index 0c35c3b..15832ce 100644
--- a/README.md
+++ b/README.md
@@ -31,7 +31,7 @@ MimiClaw turns a tiny ESP32-S3 board into a personal AI assistant. Plug it into
 
 ![](assets/mimiclaw.png)
 
-You send a message on Telegram. The ESP32-S3 picks it up over WiFi, feeds it into an agent loop — the LLM thinks, calls tools, reads memory — and sends the reply back. Supports both **Anthropic (Claude)** and **OpenAI (GPT)** as providers, switchable at runtime. Everything runs on a single $5 chip with all your data stored locally on flash.
+You send a message on Telegram. The ESP32-S3 picks it up over WiFi, feeds it into an agent loop — the LLM thinks, calls tools, reads memory — and sends the reply back. Supports **Anthropic (Claude)**, **OpenAI (GPT)**, **SiliconFlow** (Chinese LLM providers), and **Volcengine** (ByteDance Doubao models) as providers, switchable at runtime. Everything runs on a single $5 chip with all your data stored locally on flash.
 
 ## Quick Start
 
@@ -40,7 +40,7 @@ You send a message on Telegram. The ESP32-S3 picks it up over WiFi, feeds it int
 - An **ESP32-S3 dev board** with 16 MB flash and 8 MB PSRAM (e.g. Xiaozhi AI board, ~$10)
 - A **USB Type-C cable**
 - A **Telegram bot token** — talk to [@BotFather](https://t.me/BotFather) on Telegram to create one
-- An **Anthropic API key** — from [console.anthropic.com](https://console.anthropic.com), or an **OpenAI API key** — from [platform.openai.com](https://platform.openai.com)
+- An **Anthropic API key** — from [console.anthropic.com](https://console.anthropic.com), or an **OpenAI API key** — from [platform.openai.com](https://platform.openai.com), or a **SiliconFlow API key** — from [siliconflow.cn](https://siliconflow.cn), or a **Volcengine API key** — from [volcengine.com](https://volcengine.com)
 
 ### Install
 
@@ -128,11 +128,19 @@ Edit `main/mimi_secrets.h`:
 #define MIMI_SECRET_WIFI_PASS       "YourWiFiPassword"
 #define MIMI_SECRET_TG_TOKEN        "123456:ABC-DEF1234ghIkl-zyx57W2v1u123ew11"
 #define MIMI_SECRET_API_KEY         "sk-ant-api03-xxxxx"
-#define MIMI_SECRET_MODEL_PROVIDER  "anthropic"     // "anthropic" or "openai"
+#define MIMI_SECRET_MODEL_PROVIDER  "anthropic"     // "anthropic", "openai", "siliconflow", or "volcengine"
 #define MIMI_SECRET_SEARCH_KEY      ""              // optional: Brave Search API key
 #define MIMI_SECRET_TAVILY_KEY      ""              // optional: Tavily API key (preferred)
 #define MIMI_SECRET_PROXY_HOST      ""              // optional: e.g. "10.0.0.1"
 #define MIMI_SECRET_PROXY_PORT      ""              // optional: e.g. "7897"
+
+/* Optional: SiliconFlow API (OpenAI-compatible) */
+#define MIMI_SECRET_SILICONFLOW_API_KEY   ""
+#define MIMI_SECRET_SILICONFLOW_BASE_URL  "https://api.siliconflow.cn/v1"
+
+/* Optional: Volcengine API (OpenAI-compatible, ByteDance Doubao models) */
+#define MIMI_SECRET_VOLCENGINE_API_KEY    ""
+#define MIMI_SECRET_VOLCENGINE_BASE_URL   "https://ark.cn-beijing.volces.com/api/v3"
 ```
 
 Then build and flash:
@@ -169,12 +177,16 @@ Connect via serial to configure or debug. **Config commands** let you change set
 mimi> wifi_set MySSID MyPassword   # change WiFi network
 mimi> set_tg_token 123456:ABC...   # change Telegram bot token
 mimi> set_api_key sk-ant-api03-... # change API key (Anthropic or OpenAI)
-mimi> set_model_provider openai    # switch provider (anthropic|openai)
+mimi> set_model_provider openai    # switch provider (anthropic|openai|siliconflow|volcengine)
 mimi> set_model gpt-4o             # change LLM model
 mimi> set_proxy 127.0.0.1 7897  # set HTTP proxy
 mimi> clear_proxy                  # remove proxy
 mimi> set_search_key BSA...        # set Brave Search API key
 mimi> set_tavily_key tvly-...      # set Tavily API key (preferred)
+mimi> set_siliconflow_key sk-...   # set SiliconFlow API key
+mimi> set_siliconflow_url https://api.siliconflow.cn/v1  # set SiliconFlow Base URL
+mimi> set_volcengine_key sk-...    # set Volcengine API key
+mimi> set_volcengine_url https://ark.cn-beijing.volces.com/api/v3  # set Volcengine Base URL
 mimi> config_show                  # show all config (masked)
 mimi> config_reset                 # clear NVS, revert to build-time defaults
 ```
@@ -250,7 +262,7 @@ MimiClaw stores everything as plain text files you can read and edit:
 
 ## Tools
 
-MimiClaw supports tool calling for both Anthropic and OpenAI — the LLM can call tools during a conversation and loop until the task is done (ReAct pattern).
+MimiClaw supports tool calling for all LLM providers — the LLM can call tools during a conversation and loop until the task is done (ReAct pattern). Supported providers include Anthropic (Claude), OpenAI (GPT), SiliconFlow, and Volcengine.
 
 | Tool | Description |
 |------|-------------|
@@ -280,7 +292,7 @@ This turns MimiClaw into a proactive assistant — write tasks to `HEARTBEAT.md`
 - **OTA updates** — flash new firmware over WiFi, no USB needed
 - **Dual-core** — network I/O and AI processing run on separate CPU cores
 - **HTTP proxy** — CONNECT tunnel support for restricted networks
-- **Multi-provider** — supports both Anthropic (Claude) and OpenAI (GPT), switchable at runtime
+- **Multi-provider** — supports Anthropic (Claude), OpenAI (GPT), SiliconFlow, and Volcengine, switchable at runtime
 - **Cron scheduler** — the AI can schedule its own recurring and one-shot tasks, persisted across reboots
 - **Heartbeat** — periodically checks a task file and prompts the AI to act autonomously
 - **Tool use** — ReAct agent loop with tool calling for both providers
diff --git a/main/cli/serial_cli.c b/main/cli/serial_cli.c
index a77fe28..b86ae03 100644
--- a/main/cli/serial_cli.c
+++ b/main/cli/serial_cli.c
@@ -171,6 +171,80 @@ static int cmd_set_model_provider(int argc, char **argv)
     return 0;
 }
 
+/* --- set_siliconflow_key command --- */
+static struct {
+    struct arg_str *key;
+    struct arg_end *end;
+} siliconflow_key_args;
+
+static int cmd_set_siliconflow_key(int argc, char **argv)
+{
+    int nerrors = arg_parse(argc, argv, (void **)&siliconflow_key_args);
+    if (nerrors != 0) {
+        arg_print_errors(stderr, siliconflow_key_args.end, argv[0]);
+        return 1;
+    }
+    llm_set_api_key(siliconflow_key_args.key->sval[0]);
+    llm_provider_set_api_key("siliconflow", siliconflow_key_args.key->sval[0]);
+    printf("SiliconFlow API key saved.\n");
+    return 0;
+}
+
+/* --- set_siliconflow_url command --- */
+static struct {
+    struct arg_str *url;
+    struct arg_end *end;
+} siliconflow_url_args;
+
+static int cmd_set_siliconflow_url(int argc, char **argv)
+{
+    int nerrors = arg_parse(argc, argv, (void **)&siliconflow_url_args);
+    if (nerrors != 0) {
+        arg_print_errors(stderr, siliconflow_url_args.end, argv[0]);
+        return 1;
+    }
+    llm_set_base_url("siliconflow", siliconflow_url_args.url->sval[0]);
+    printf("SiliconFlow Base URL saved.\n");
+    return 0;
+}
+
+/* --- set_volcengine_key command --- */
+static struct {
+    struct arg_str *key;
+    struct arg_end *end;
+} volcengine_key_args;
+
+static int cmd_set_volcengine_key(int argc, char **argv)
+{
+    int nerrors = arg_parse(argc, argv, (void **)&volcengine_key_args);
+    if (nerrors != 0) {
+        arg_print_errors(stderr, volcengine_key_args.end, argv[0]);
+        return 1;
+    }
+    llm_set_api_key(volcengine_key_args.key->sval[0]);
+    llm_provider_set_api_key("volcengine", volcengine_key_args.key->sval[0]);
+    printf("Volcengine API key saved.\n");
+    return 0;
+}
+
+/* --- set_volcengine_url command --- */
+static struct {
+    struct arg_str *url;
+    struct arg_end *end;
+} volcengine_url_args;
+
+static int cmd_set_volcengine_url(int argc, char **argv)
+{
+    int nerrors = arg_parse(argc, argv, (void **)&volcengine_url_args);
+    if (nerrors != 0) {
+        arg_print_errors(stderr, volcengine_url_args.end, argv[0]);
+        return 1;
+    }
+    llm_set_base_url("volcengine", volcengine_url_args.url->sval[0]);
+    printf("Volcengine Base URL saved.\n");
+    return 0;
+}
+
 /* --- memory_read command --- */
 static int cmd_memory_read(int argc, char **argv)
 {
@@ -563,6 +637,18 @@ static int cmd_config_show(int argc, char **argv)
     print_config("API Key",    MIMI_NVS_LLM,    MIMI_NVS_KEY_API_KEY,  MIMI_SECRET_API_KEY,    true);
     print_config("Model",      MIMI_NVS_LLM,    MIMI_NVS_KEY_MODEL,    MIMI_SECRET_MODEL,      false);
     print_config("Provider",   MIMI_NVS_LLM,    MIMI_NVS_KEY_PROVIDER, MIMI_SECRET_MODEL_PROVIDER, false);
+    
+    /* Provider-specific configurations */
+    printf("  --- Provider-specific configs ---\n");
+    print_config("Anthropic Key", MIMI_NVS_LLM, MIMI_NVS_KEY_ANTHROPIC_API_KEY, "", true);
+    print_config("Anthropic URL", MIMI_NVS_LLM, MIMI_NVS_KEY_ANTHROPIC_BASE_URL, "", false);
+    print_config("OpenAI Key", MIMI_NVS_LLM, MIMI_NVS_KEY_OPENAI_API_KEY, "", true);
+    print_config("OpenAI URL", MIMI_NVS_LLM, MIMI_NVS_KEY_OPENAI_BASE_URL, "", false);
+    print_config("SiliconFlow Key", MIMI_NVS_LLM, MIMI_NVS_KEY_SILICONFLOW_API_KEY, "", true);
+    print_config("SiliconFlow URL", MIMI_NVS_LLM, MIMI_NVS_KEY_SILICONFLOW_BASE_URL, "", false);
+    print_config("Volcengine Key", MIMI_NVS_LLM, MIMI_NVS_KEY_VOLCENGINE_API_KEY, "", true);
+    print_config("Volcengine URL", MIMI_NVS_LLM, MIMI_NVS_KEY_VOLCENGINE_BASE_URL, "", false);
+    
     print_config("Proxy Host", MIMI_NVS_PROXY,  MIMI_NVS_KEY_PROXY_HOST, MIMI_SECRET_PROXY_HOST, false);
     print_config_u16("Proxy Port", MIMI_NVS_PROXY, MIMI_NVS_KEY_PROXY_PORT, MIMI_SECRET_PROXY_PORT);
     print_config("Search Key", MIMI_NVS_SEARCH, MIMI_NVS_KEY_API_KEY,  MIMI_SECRET_SEARCH_KEY, true);
@@ -887,7 +973,7 @@ esp_err_t serial_cli_init(void)
     esp_console_cmd_register(&model_cmd);
 
     /* set_model_provider */
-    provider_args.provider = arg_str1(NULL, NULL, "<provider>", "Model provider (anthropic|openai)");
+    provider_args.provider = arg_str1(NULL, NULL, "<provider>", "Model provider (anthropic|openai|siliconflow|volcengine)");
     provider_args.end = arg_end(1);
     esp_console_cmd_t provider_cmd = {
         .command = "set_model_provider",
@@ -897,6 +983,50 @@ esp_err_t serial_cli_init(void)
     };
     esp_console_cmd_register(&provider_cmd);
 
+    /* set_siliconflow_key */
+    siliconflow_key_args.key = arg_str1(NULL, NULL, "<key>", "SiliconFlow API key");
+    siliconflow_key_args.end = arg_end(1);
+    esp_console_cmd_t siliconflow_key_cmd = {
+        .command = "set_siliconflow_key",
+        .help = "Set SiliconFlow API key",
+        .func = &cmd_set_siliconflow_key,
+        .argtable = &siliconflow_key_args,
+    };
+    esp_console_cmd_register(&siliconflow_key_cmd);
+
+    /* set_siliconflow_url */
+    siliconflow_url_args.url = arg_str1(NULL, NULL, "<url>", "SiliconFlow Base URL");
+    siliconflow_url_args.end = arg_end(1);
+    esp_console_cmd_t siliconflow_url_cmd = {
+        .command = "set_siliconflow_url",
+        .help = "Set SiliconFlow Base URL",
+        .func = &cmd_set_siliconflow_url,
+        .argtable = &siliconflow_url_args,
+    };
+    esp_console_cmd_register(&siliconflow_url_cmd);
+
+    /* set_volcengine_key */
+    volcengine_key_args.key = arg_str1(NULL, NULL, "<key>", "Volcengine API key");
+    volcengine_key_args.end = arg_end(1);
+    esp_console_cmd_t volcengine_key_cmd = {
+        .command = "set_volcengine_key",
+        .help = "Set Volcengine API key",
+        .func = &cmd_set_volcengine_key,
+        .argtable = &volcengine_key_args,
+    };
+    esp_console_cmd_register(&volcengine_key_cmd);
+
+    /* set_volcengine_url */
+    volcengine_url_args.url = arg_str1(NULL, NULL, "<url>", "Volcengine Base URL");
+    volcengine_url_args.end = arg_end(1);
+    esp_console_cmd_t volcengine_url_cmd = {
+        .command = "set_volcengine_url",
+        .help = "Set Volcengine Base URL",
+        .func = &cmd_set_volcengine_url,
+        .argtable = &volcengine_url_args,
+    };
+    esp_console_cmd_register(&volcengine_url_cmd);
+
     /* skill_list */
     esp_console_cmd_t skill_list_cmd = {
         .command = "skill_list",
diff --git a/main/llm/llm_provider.c b/main/llm/llm_provider.c
new file mode 100644
index 0000000..10d78aa
--- /dev/null
+++ b/main/llm/llm_provider.c
@@ -0,0 +1,298 @@
+#include "llm_provider.h"
+#include "mimi_config.h"
+#include "nvs.h"
+
+#include <string.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include "esp_log.h"
+#include "esp_err.h"
+
+static const char *TAG = "llm_provider";
+
+#define LLM_API_KEY_MAX_LEN 320
+
+/* Provider registry - all supported providers */
+const llm_provider_config_t llm_providers[] = {
+    {
+        .name = "anthropic",
+        .default_api_url = MIMI_LLM_API_URL,
+        .default_host = "api.anthropic.com",
+        .default_path = "/v1/messages",
+        .is_openai_compatible = false,
+    },
+    {
+        .name = "openai",
+        .default_api_url = MIMI_OPENAI_API_URL,
+        .default_host = "api.openai.com",
+        .default_path = "/v1/chat/completions",
+        .is_openai_compatible = true,
+    },
+    {
+        .name = "siliconflow",
+        .default_api_url = MIMI_SILICONFLOW_API_URL,
+        .default_host = "api.siliconflow.cn",
+        .default_path = "/v1/chat/completions",
+        .is_openai_compatible = true,
+    },
+    {
+        .name = "volcengine",
+        .default_api_url = MIMI_VOLCENGINE_API_URL,
+        .default_host = "ark.cn-beijing.volces.com",
+        .default_path = "/v1/chat/completions",
+        .is_openai_compatible = true,
+    },
+};
+
+const int llm_provider_count = sizeof(llm_providers) / sizeof(llm_providers[0]);
+
+/* Current provider state */
+static const llm_provider_config_t *s_current_provider = &llm_providers[0]; /* Default to anthropic */
+static char s_api_key[LLM_API_KEY_MAX_LEN] = {0};
+static char s_base_url[256] = {0};
+
+/* Helper function to get NVS key for provider API key */
+static const char *get_provider_api_key_nvs_key(const char *provider_name) {
+    if (strcmp(provider_name, "anthropic") == 0) return MIMI_NVS_KEY_ANTHROPIC_API_KEY;
+    if (strcmp(provider_name, "openai") == 0) return MIMI_NVS_KEY_OPENAI_API_KEY;
+    if (strcmp(provider_name, "siliconflow") == 0) return MIMI_NVS_KEY_SILICONFLOW_API_KEY;
+    if (strcmp(provider_name, "volcengine") == 0) return MIMI_NVS_KEY_VOLCENGINE_API_KEY;
+    return NULL;
+}
+
+/* Helper function to get NVS key for provider Base URL */
+static const char *get_provider_base_url_nvs_key(const char *provider_name) {
+    if (strcmp(provider_name, "anthropic") == 0) return MIMI_NVS_KEY_ANTHROPIC_BASE_URL;
+    if (strcmp(provider_name, "openai") == 0) return MIMI_NVS_KEY_OPENAI_BASE_URL;
+    if (strcmp(provider_name, "siliconflow") == 0) return MIMI_NVS_KEY_SILICONFLOW_BASE_URL;
+    if (strcmp(provider_name, "volcengine") == 0) return MIMI_NVS_KEY_VOLCENGINE_BASE_URL;
+    return NULL;
+}
+
+/* Find provider configuration by name */
+const llm_provider_config_t *llm_provider_find(const char *name) {
+    if (!name) return NULL;
+    
+    for (int i = 0; i < llm_provider_count; i++) {
+        if (strcmp(llm_providers[i].name, name) == 0) {
+            return &llm_providers[i];
+        }
+    }
+    return NULL;
+}
+
+/* Get current provider configuration */
+const llm_provider_config_t *llm_provider_current(void) {
+    return s_current_provider;
+}
+
+/* Set current provider by name */
+void llm_provider_set_current(const char *name) {
+    const llm_provider_config_t *provider = llm_provider_find(name);
+    if (provider) {
+        s_current_provider = provider;
+        /* Load provider-specific configuration */
+        llm_provider_init();
+        ESP_LOGI(TAG, "Current provider set to: %s", name);
+    } else {
+        ESP_LOGW(TAG, "Unknown provider: %s", name);
+    }
+}
+
+/* Get current provider name */
+const char *llm_provider_current_name(void) {
+    return s_current_provider ? s_current_provider->name : "unknown";
+}
+
+/* Check if current provider is OpenAI-compatible */
+bool llm_provider_is_openai_compatible(void) {
+    return s_current_provider ? s_current_provider->is_openai_compatible : false;
+}
+
+/* Get API URL for current provider (with dynamic Base URL support) */
+const char *llm_provider_api_url(void) {
+    if (s_base_url[0] != '\0') {
+        /* Use configured Base URL */
+        static char full_url[512];
+        snprintf(full_url, sizeof(full_url), "%s%s", s_base_url, s_current_provider->default_path);
+        return full_url;
+    }
+    /* Use default API URL */
+    return s_current_provider ? s_current_provider->default_api_url : "";
+}
+
+/* Get hostname for current provider */
+const char *llm_provider_host(void) {
+    if (s_base_url[0] != '\0') {
+        /* Extract hostname from Base URL */
+        static char hostname[256];
+        const char *start = s_base_url;
+        if (strncmp(start, "https://", 8) == 0) start += 8;
+        else if (strncmp(start, "http://", 7) == 0) start += 7;
+        
+        const char *end = strchr(start, '/');
+        if (end) {
+            size_t len = end - start;
+            if (len < sizeof(hostname)) {
+                memcpy(hostname, start, len);
+                hostname[len] = '\0';
+                return hostname;
+            }
+        }
+        /* No path, copy whole string */
+        strncpy(hostname, start, sizeof(hostname) - 1);
+        hostname[sizeof(hostname) - 1] = '\0';
+        return hostname;
+    }
+    /* Use default host */
+    return s_current_provider ? s_current_provider->default_host : "";
+}
+
+/* Get API path for current provider */
+const char *llm_provider_path(void) {
+    return s_current_provider ? s_current_provider->default_path : "";
+}
+
+/* Provider-specific API key management */
+void llm_provider_set_api_key(const char *provider_name, const char *api_key) {
+    if (!provider_name || !api_key) return;
+    
+    const char *nvs_key = get_provider_api_key_nvs_key(provider_name);
+    if (!nvs_key) {
+        ESP_LOGW(TAG, "No NVS key for provider: %s", provider_name);
+        return;
+    }
+    
+    nvs_handle_t nvs;
+    if (nvs_open(MIMI_NVS_LLM, NVS_READWRITE, &nvs) == ESP_OK) {
+        nvs_set_str(nvs, nvs_key, api_key);
+        nvs_commit(nvs);
+        nvs_close(nvs);
+        ESP_LOGI(TAG, "API key saved for provider: %s", provider_name);
+    }
+    
+    /* If this is the current provider, update in-memory key */
+    if (strcmp(provider_name, s_current_provider->name) == 0) {
+        strncpy(s_api_key, api_key, sizeof(s_api_key) - 1);
+        s_api_key[sizeof(s_api_key) - 1] = '\0';
+    }
+}
+
+/* Provider-specific Base URL management */
+void llm_provider_set_base_url(const char *provider_name, const char *base_url) {
+    if (!provider_name || !base_url) return;
+    
+    const char *nvs_key = get_provider_base_url_nvs_key(provider_name);
+    if (!nvs_key) {
+        ESP_LOGW(TAG, "No NVS key for provider: %s", provider_name);
+        return;
+    }
+    
+    nvs_handle_t nvs;
+    if (nvs_open(MIMI_NVS_LLM, NVS_READWRITE, &nvs) == ESP_OK) {
+        nvs_set_str(nvs, nvs_key, base_url);
+        nvs_commit(nvs);
+        nvs_close(nvs);
+        ESP_LOGI(TAG, "Base URL saved for provider: %s", provider_name);
+    }
+    
+    /* If this is the current provider, update in-memory Base URL */
+    if (strcmp(provider_name, s_current_provider->name) == 0) {
+        strncpy(s_base_url, base_url, sizeof(s_base_url) - 1);
+        s_base_url[sizeof(s_base_url) - 1] = '\0';
+    }
+}
+
+/* Get API key for a provider */
+const char *llm_provider_get_api_key(const char *provider_name) {
+    if (!provider_name) return NULL;
+    
+    /* If this is the current provider, return in-memory key */
+    if (strcmp(provider_name, s_current_provider->name) == 0) {
+        return s_api_key;
+    }
+    
+    /* Otherwise, load from NVS */
+    static char key_buffer[LLM_API_KEY_MAX_LEN] = {0};
+    const char *nvs_key = get_provider_api_key_nvs_key(provider_name);
+    if (!nvs_key) return NULL;
+    
+    nvs_handle_t nvs;
+    if (nvs_open(MIMI_NVS_LLM, NVS_READONLY, &nvs) == ESP_OK) {
+        size_t len = sizeof(key_buffer);
+        if (nvs_get_str(nvs, nvs_key, key_buffer, &len) == ESP_OK && key_buffer[0]) {
+            nvs_close(nvs);
+            return key_buffer;
+        }
+        nvs_close(nvs);
+    }
+    return NULL;
+}
+
+/* Get Base URL for a provider */
+const char *llm_provider_get_base_url(const char *provider_name) {
+    if (!provider_name) return NULL;
+    
+    /* If this is the current provider, return in-memory Base URL */
+    if (strcmp(provider_name, s_current_provider->name) == 0) {
+        return s_base_url[0] ? s_base_url : NULL;
+    }
+    
+    /* Otherwise, load from NVS */
+    static char url_buffer[256] = {0};
+    const char *nvs_key = get_provider_base_url_nvs_key(provider_name);
+    if (!nvs_key) return NULL;
+    
+    nvs_handle_t nvs;
+    if (nvs_open(MIMI_NVS_LLM, NVS_READONLY, &nvs) == ESP_OK) {
+        size_t len = sizeof(url_buffer);
+        if (nvs_get_str(nvs, nvs_key, url_buffer, &len) == ESP_OK && url_buffer[0]) {
+            nvs_close(nvs);
+            return url_buffer;
+        }
+        nvs_close(nvs);
+    }
+    return NULL;
+}
+
+/* Initialize provider system (load from NVS) */
+void llm_provider_init(void) {
+    /* Load API key for current provider */
+    const char *api_key = llm_provider_get_api_key(s_current_provider->name);
+    if (api_key) {
+        strncpy(s_api_key, api_key, sizeof(s_api_key) - 1);
+        s_api_key[sizeof(s_api_key) - 1] = '\0';
+    } else {
+        s_api_key[0] = '\0';
+    }
+    
+    /* Load Base URL for current provider */
+    const char *base_url = llm_provider_get_base_url(s_current_provider->name);
+    if (base_url) {
+        strncpy(s_base_url, base_url, sizeof(s_base_url) - 1);
+        s_base_url[sizeof(s_base_url) - 1] = '\0';
+    } else {
+        s_base_url[0] = '\0';
+    }
+    
+    ESP_LOGI(TAG, "Provider initialized: %s (API key: %s, Base URL: %s)",
+             s_current_provider->name,
+             s_api_key[0] ? "set" : "not set",
+             s_base_url[0] ? s_base_url : "default");
+}
+
+/* Save provider configuration to NVS */
+void llm_provider_save_config(const char *provider_name) {
+    /* This function is intentionally left empty - 
+       individual set functions already save to NVS */
+}
+
+/* Common authentication header setup for OpenAI-compatible providers */
+void llm_provider_set_auth_headers(esp_http_client_handle_t client, const char *api_key) {
+    if (!client || !api_key) return;
+    
+    /* All OpenAI-compatible providers use Bearer token authentication */
+    char auth[LLM_API_KEY_MAX_LEN + 16];
+    snprintf(auth, sizeof(auth), "Bearer %s", api_key);
+    esp_http_client_set_header(client, "Authorization", auth);
+}
\ No newline at end of file
diff --git a/main/llm/llm_provider.h b/main/llm/llm_provider.h
new file mode 100644
index 0000000..713167a
--- /dev/null
+++ b/main/llm/llm_provider.h
@@ -0,0 +1,68 @@
+#pragma once
+
+#include <stdbool.h>
+#include <stddef.h>
+#include "esp_err.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* Forward declaration for esp_http_client_handle_t */
+typedef struct esp_http_client* esp_http_client_handle_t;
+
+/* Provider configuration structure */
+typedef struct {
+    const char *name;                    /* Provider name, e.g., "anthropic", "openai", "siliconflow", "volcengine" */
+    const char *default_api_url;         /* Default API URL */
+    const char *default_host;            /* Default hostname */
+    const char *default_path;            /* Default API path */
+    bool is_openai_compatible;           /* Whether this provider uses OpenAI-compatible API */
+} llm_provider_config_t;
+
+/* Provider registry - all supported providers */
+extern const llm_provider_config_t llm_providers[];
+extern const int llm_provider_count;
+
+/* Find provider configuration by name */
+const llm_provider_config_t *llm_provider_find(const char *name);
+
+/* Get current provider configuration */
+const llm_provider_config_t *llm_provider_current(void);
+
+/* Set current provider by name */
+void llm_provider_set_current(const char *name);
+
+/* Get current provider name */
+const char *llm_provider_current_name(void);
+
+/* Check if current provider is OpenAI-compatible */
+bool llm_provider_is_openai_compatible(void);
+
+/* Get API URL for current provider (with dynamic Base URL support) */
+const char *llm_provider_api_url(void);
+
+/* Get hostname for current provider */
+const char *llm_provider_host(void);
+
+/* Get API path for current provider */
+const char *llm_provider_path(void);
+
+/* Provider-specific API key and Base URL management */
+void llm_provider_set_api_key(const char *provider_name, const char *api_key);
+void llm_provider_set_base_url(const char *provider_name, const char *base_url);
+const char *llm_provider_get_api_key(const char *provider_name);
+const char *llm_provider_get_base_url(const char *provider_name);
+
+/* Initialize provider system (load from NVS) */
+void llm_provider_init(void);
+
+/* Save provider configuration to NVS */
+void llm_provider_save_config(const char *provider_name);
+
+/* Common authentication header setup for OpenAI-compatible providers */
+void llm_provider_set_auth_headers(esp_http_client_handle_t client, const char *api_key);
+
+#ifdef __cplusplus
+}
+#endif
\ No newline at end of file
diff --git a/main/llm/llm_proxy.c b/main/llm/llm_proxy.c
index c6fa1b8..09f73cc 100644
--- a/main/llm/llm_proxy.c
+++ b/main/llm/llm_proxy.c
@@ -1,4 +1,5 @@
 #include "llm_proxy.h"
+#include "llm_provider.h"
 #include "mimi_config.h"
 #include "proxy/http_proxy.h"
 
@@ -184,28 +185,31 @@ static esp_err_t http_event_handler(esp_http_client_event_t *evt)
 
 static bool provider_is_openai(void)
 {
-    return strcmp(s_provider, "openai") == 0;
+    return llm_provider_is_openai_compatible();
 }
 
 static const char *llm_api_url(void)
 {
-    return provider_is_openai() ? MIMI_OPENAI_API_URL : MIMI_LLM_API_URL;
+    return llm_provider_api_url();
 }
 
 static const char *llm_api_host(void)
 {
-    return provider_is_openai() ? "api.openai.com" : "api.anthropic.com";
+    return llm_provider_host();
 }
 
 static const char *llm_api_path(void)
 {
-    return provider_is_openai() ? "/v1/chat/completions" : "/v1/messages";
+    return llm_provider_path();
 }
 
 /* ── Init ─────────────────────────────────────────────────────── */
 
 esp_err_t llm_proxy_init(void)
 {
+    /* Initialize provider system */
+    llm_provider_init();
+    
     /* Start with build-time defaults */
     if (MIMI_SECRET_API_KEY[0] != '\0') {
         safe_copy(s_api_key, sizeof(s_api_key), MIMI_SECRET_API_KEY);
@@ -215,6 +219,8 @@ esp_err_t llm_proxy_init(void)
     }
     if (MIMI_SECRET_MODEL_PROVIDER[0] != '\0') {
         safe_copy(s_provider, sizeof(s_provider), MIMI_SECRET_MODEL_PROVIDER);
+        /* Set current provider based on build-time default */
+        llm_provider_set_current(s_provider);
     }
 
     /* NVS overrides take highest priority (set via CLI) */
@@ -234,10 +240,18 @@ esp_err_t llm_proxy_init(void)
         len = sizeof(provider_tmp);
         if (nvs_get_str(nvs, MIMI_NVS_KEY_PROVIDER, provider_tmp, &len) == ESP_OK && provider_tmp[0]) {
             safe_copy(s_provider, sizeof(s_provider), provider_tmp);
+            /* Set current provider based on NVS override */
+            llm_provider_set_current(s_provider);
         }
         nvs_close(nvs);
     }
 
+    /* Load provider-specific API key if available */
+    const char *provider_api_key = llm_provider_get_api_key(s_provider);
+    if (provider_api_key && provider_api_key[0]) {
+        safe_copy(s_api_key, sizeof(s_api_key), provider_api_key);
+    }
+
     if (s_api_key[0]) {
         ESP_LOGI(TAG, "LLM proxy initialized (provider: %s, model: %s)", s_provider, s_model);
     } else {
@@ -265,13 +279,15 @@ static esp_err_t llm_http_direct(const char *post_data, resp_buf_t *rb, int *out
 
     esp_http_client_set_method(client, HTTP_METHOD_POST);
     esp_http_client_set_header(client, "Content-Type", "application/json");
-    if (provider_is_openai()) {
+    
+    /* Use provider-specific authentication */
+    if (llm_provider_is_openai_compatible()) {
+        /* OpenAI-compatible providers use Bearer token authentication */
         if (s_api_key[0]) {
-            char auth[LLM_API_KEY_MAX_LEN + 16];
-            snprintf(auth, sizeof(auth), "Bearer %s", s_api_key);
-            esp_http_client_set_header(client, "Authorization", auth);
+            llm_provider_set_auth_headers(client, s_api_key);
         }
     } else {
+        /* Anthropic uses x-api-key authentication */
         esp_http_client_set_header(client, "x-api-key", s_api_key);
         esp_http_client_set_header(client, "anthropic-version", MIMI_LLM_API_VERSION);
     }
@@ -293,7 +309,9 @@ static esp_err_t llm_http_via_proxy(const char *post_data, resp_buf_t *rb, int *
     int body_len = strlen(post_data);
     char header[1024];
     int hlen = 0;
-    if (provider_is_openai()) {
+    
+    if (llm_provider_is_openai_compatible()) {
+        /* OpenAI-compatible providers use Bearer token authentication */
         hlen = snprintf(header, sizeof(header),
             "POST %s HTTP/1.1\r\n"
             "Host: %s\r\n"
@@ -303,6 +321,7 @@ static esp_err_t llm_http_via_proxy(const char *post_data, resp_buf_t *rb, int *
             "Connection: close\r\n\r\n",
             llm_api_path(), llm_api_host(), s_api_key, body_len);
     } else {
+        /* Anthropic uses x-api-key authentication */
         hlen = snprintf(header, sizeof(header),
             "POST %s HTTP/1.1\r\n"
             "Host: %s\r\n"
@@ -559,13 +578,13 @@ esp_err_t llm_chat_tools(const char *system_prompt,
     /* Build request body (non-streaming) */
     cJSON *body = cJSON_CreateObject();
     cJSON_AddStringToObject(body, "model", s_model);
-    if (provider_is_openai()) {
+    if (llm_provider_is_openai_compatible()) {
         cJSON_AddNumberToObject(body, "max_completion_tokens", MIMI_LLM_MAX_TOKENS);
     } else {
         cJSON_AddNumberToObject(body, "max_tokens", MIMI_LLM_MAX_TOKENS);
     }
 
-    if (provider_is_openai()) {
+    if (llm_provider_is_openai_compatible()) {
         cJSON *openai_msgs = convert_messages_openai(system_prompt, messages);
         cJSON_AddItemToObject(body, "messages", openai_msgs);
 
@@ -635,7 +654,7 @@ esp_err_t llm_chat_tools(const char *system_prompt,
         return ESP_FAIL;
     }
 
-    if (provider_is_openai()) {
+    if (llm_provider_is_openai_compatible()) {
         cJSON *choices = cJSON_GetObjectItem(root, "choices");
         cJSON *choice0 = choices && cJSON_IsArray(choices) ? cJSON_GetArrayItem(choices, 0) : NULL;
         if (choice0) {
@@ -780,7 +799,11 @@ esp_err_t llm_set_api_key(const char *api_key)
     nvs_close(nvs);
 
     safe_copy(s_api_key, sizeof(s_api_key), api_key);
-    ESP_LOGI(TAG, "API key saved");
+    
+    /* Also save to provider-specific NVS key */
+    llm_provider_set_api_key(s_provider, api_key);
+    
+    ESP_LOGI(TAG, "API key saved for provider: %s", s_provider);
     return ESP_OK;
 }
 
@@ -806,6 +829,43 @@ esp_err_t llm_set_provider(const char *provider)
     nvs_close(nvs);
 
     safe_copy(s_provider, sizeof(s_provider), provider);
+    
+    /* Update current provider in the provider system */
+    llm_provider_set_current(provider);
+    
+    /* Load provider-specific API key if available */
+    const char *provider_api_key = llm_provider_get_api_key(provider);
+    if (provider_api_key && provider_api_key[0]) {
+        safe_copy(s_api_key, sizeof(s_api_key), provider_api_key);
+    } else {
+        s_api_key[0] = '\0';
+    }
+    
     ESP_LOGI(TAG, "Provider set to: %s", s_provider);
     return ESP_OK;
 }
+
+esp_err_t llm_set_base_url(const char *provider, const char *base_url)
+{
+    if (!provider || !base_url) return ESP_ERR_INVALID_ARG;
+    
+    /* Save to provider-specific NVS key */
+    llm_provider_set_base_url(provider, base_url);
+    
+    /* If this is the current provider, update in-memory Base URL */
+    if (strcmp(provider, s_provider) == 0) {
+        /* Reload provider configuration to pick up new Base URL */
+        llm_provider_init();
+    }
+    
+    ESP_LOGI(TAG, "Base URL set for provider: %s", provider);
+    return ESP_OK;
+}
+
+const char *llm_get_base_url(const char *provider)
+{
+    if (!provider) return NULL;
+    
+    /* Get Base URL from provider system */
+    return llm_provider_get_base_url(provider);
+}
diff --git a/main/llm/llm_proxy.h b/main/llm/llm_proxy.h
index b667f62..516bbbf 100644
--- a/main/llm/llm_proxy.h
+++ b/main/llm/llm_proxy.h
@@ -27,6 +27,16 @@ esp_err_t llm_set_provider(const char *provider);
  */
 esp_err_t llm_set_model(const char *model);
 
+/**
+ * Save the Base URL for a provider to NVS.
+ */
+esp_err_t llm_set_base_url(const char *provider, const char *base_url);
+
+/**
+ * Get the Base URL for a provider from NVS.
+ */
+const char *llm_get_base_url(const char *provider);
+
 /* ── Tool Use Support ──────────────────────────────────────────── */
 
 typedef struct {
diff --git a/main/mimi_config.h b/main/mimi_config.h
index f205c54..55eb7d0 100644
--- a/main/mimi_config.h
+++ b/main/mimi_config.h
@@ -88,6 +88,8 @@
 #define MIMI_LLM_MAX_TOKENS          4096
 #define MIMI_LLM_API_URL             "https://api.anthropic.com/v1/messages"
 #define MIMI_OPENAI_API_URL          "https://api.openai.com/v1/chat/completions"
+#define MIMI_SILICONFLOW_API_URL     "https://api.siliconflow.cn/v1/chat/completions"
+#define MIMI_VOLCENGINE_API_URL      "https://ark.cn-beijing.volces.com/api/v3/chat/completions"
 #define MIMI_LLM_API_VERSION         "2023-06-01"
 #define MIMI_LLM_STREAM_BUF_SIZE     (32 * 1024)
 #define MIMI_LLM_LOG_VERBOSE_PAYLOAD 0
@@ -154,6 +156,16 @@
 #define MIMI_NVS_KEY_PROXY_PORT      "port"
 #define MIMI_NVS_KEY_PROXY_TYPE      "proxy_type"
 
+/* Provider-specific NVS Keys */
+#define MIMI_NVS_KEY_ANTHROPIC_API_KEY   "anthropic_api_key"
+#define MIMI_NVS_KEY_ANTHROPIC_BASE_URL  "anthropic_base_url"
+#define MIMI_NVS_KEY_OPENAI_API_KEY      "openai_api_key"
+#define MIMI_NVS_KEY_OPENAI_BASE_URL     "openai_base_url"
+#define MIMI_NVS_KEY_SILICONFLOW_API_KEY   "siliconflow_api_key"
+#define MIMI_NVS_KEY_SILICONFLOW_BASE_URL  "siliconflow_base_url"
+#define MIMI_NVS_KEY_VOLCENGINE_API_KEY    "volcengine_api_key"
+#define MIMI_NVS_KEY_VOLCENGINE_BASE_URL   "volcengine_base_url"
+
 /* WiFi Onboarding (Captive Portal) */
 #define MIMI_ONBOARD_AP_PREFIX    "MimiClaw-"
 #define MIMI_ONBOARD_AP_PASS      ""          /* open network */
diff --git a/main/mimi_secrets.h.example b/main/mimi_secrets.h.example
index ecebf54..c5050f4 100644
--- a/main/mimi_secrets.h.example
+++ b/main/mimi_secrets.h.example
@@ -26,6 +26,14 @@
 #define MIMI_SECRET_MODEL           ""
 #define MIMI_SECRET_MODEL_PROVIDER  "anthropic"
 
+/* SiliconFlow API (OpenAI-compatible) */
+#define MIMI_SECRET_SILICONFLOW_API_KEY   ""
+#define MIMI_SECRET_SILICONFLOW_BASE_URL  "https://api.siliconflow.cn/v1"
+
+/* Volcengine API (OpenAI-compatible, ByteDance Doubao models) */
+#define MIMI_SECRET_VOLCENGINE_API_KEY    ""
+#define MIMI_SECRET_VOLCENGINE_BASE_URL   "https://ark.cn-beijing.volces.com/api/v3"
+
 /* HTTP Proxy (leave empty or set both) */
 #define MIMI_SECRET_PROXY_HOST      ""
 #define MIMI_SECRET_PROXY_PORT      ""